Mechanism of IKK activation by the Kaposi's sarcoma ... - UCL Discovery

Mechanism of IKK activation by the Kaposi's sarcoma-associated herpesvirus protein vFLIP and its cellular homologues

Mehdi Baratchian

Thesis submitted to the University College London for the degree of Doctor of Philosophy 2014

Division of Infection and Immunity University College London

DECLARATION

I, Mehdi Baratchian, confirm the work presented in this thesis is my own. Where information has been derived from other sources, I confirm that this has been indicated in the thesis. No part of this thesis has been or is currently being submitted for any other degree in this university or elsewhere.

‫د ن ‪  6‬ا‪ 5‬و دن‬ ‫  رد د را ن د ران‬ ‫‪7‬‬

‫‪3‬‬

ABSTRACT

Activation of the NF-κB pathway is linked to cancer development and progression. Kaposi’s sarcoma-associated herpesvirus (KSHV/HHV8) encodes vFLIP which binds to the NEMO/IKKγ subunit of IKK and constitutively activates NF-κB, leading to tumourigenesis. Cellular FLIPs, which share sequence homology with KSHV vFLIP and induce NF-κB activation, are upregulated in a variety of malignancies and are therefore promising targets for anti-cancer therapies. The cFLIP family consists of three splice variant isoforms (cFLIPL, cFLIPS and cFLIPR) and two proteolytic fragments (p43-FLIP and p22-FLIP). Much is known about how cFLIPs regulate apoptosis but the mechanisms by which they activate NF-κB are not well understood. Potential similarities to vFLIP-induced activation have been suggested but not investigated. Here we show that, unlike KSHV vFLIP, cFLIP variants are not found in stable complexes with NEMO and all require upstream events to mediate signalling to IKK. By mutational analysis on NEMO and protein expression knockdowns, we demonstrate that all cFLIP isoforms require the ubiquitin binding domain (UBD) of NEMO while it is redundant for vFLIP’s function. Similarly, our data reveals that TAK1 is essential for induction of IKK by cFLIP isoforms but not vFLIP. We further show that different cFLIP isoforms have different requirements for IKK activation. While cFLIPL needs LUBAC to activate NF-κB, cFLIPS and p22-FLIP require FADD and RIP1. Contrary to existing reports, our results suggest that processing of cFLIPL to p22-FLIP or p43-FLIP fragments by caspase-8 is not necessary for its IKK activation. Finally, we propose that vFLIP-mediated activation of IKK is most likely to occur through induction of multimerisation and re-orientation of the IKK complexes within higher order IKK assemblies that lead to autophosphorylation of the enzymatic subunits, IKKα and β. In conclusion, the work in this thesis provides evidence that vFLIP, cFLIPL, cFLIPS and p22-FLIP have specific and different mechanisms of inducing IKK activation. This has implications for the design of therapeutics to block pathological NF-κB activation in viral and non-viral tumours.

4

ACKNOWLEDGEMENTS

First and foremost, I am deeply grateful to Mary Collins for supervising this PhD. Her constant encouragement, guidance, patience and support from the first time I met her in Tehran has gone beyond my expectations. Without her help, this would have not been possible. It has been a privilege to be part of the Collins lab and I would like to thank all the past and present members of the lab. An especially big thank you to David Escors for training me during the early stages of my PhD and his great advice during the past four years. Christopher Bricogne’s help and friendship are highly valued legacies of our time in the UCL Cancer Institute and I will fondly remember our 8 pm discussions in the lab. I also owe many thanks to fellow colleagues: Doug, Fred, Sean, Khaled, Gary, Chris and Kam for providing a fantastic environment to work in and for many scientific and nonscientific discussions we have had over these years. A word of thanks goes to my MPhil/PhD examiners -Mahdad Noursadeghi and Pablo Rodriguez- for their advice. I also gratefully acknowledge the funding from UCL graduate school which made this PhD possible. I am eternally grateful to my parents Rafat and Zaman, my sister Sima, my brother Mohammad Reza and my nephews Sina and Taha for their love, support and belief in me. In particular, my late father who has been and always will remain my source of inspiration in life. To him, I dedicate this thesis. I am also indebted to my parents in-law, Hoda and Imad, for their unconditional love and support throughout. An enormous thank you to my cousin Mohamad Alimohamadian for everything he has done for me. You are a great friend and have always been close to me in spite of distance. Last but not least, to my best friend and wife Maha, not enough words can describe how grateful I am. You have been a constant source of strength and happiness for me and have always changed me for the better. I love you.

5

TABLE OF CONTENTS

DECLARATION .....................................................................................2 ABSTRACT..............................................................................................4 ACKNOWLEDGEMENTS ....................................................................5 TABLE OF CONTENTS .......................................................................6 LIST OF FIGURES .................................................................................9 LIST OF TABLES ................................................................................. 11 ABBREVIATIONS ............................................................................... 12 1. INTRODUCTION ...................................................................... 15 1.1 NF-κB signalling pathway: an overview ....................................................... 16 1.1.1 NF-κB related proteins ............................................................................................ 16 1.1.2 Two pathways to NF-κB ......................................................................................... 23 1.2 Mechanisms behind positive and negative regulation of IKK ..................27 1.2.1 IKK-activating kinases............................................................................................. 27 1.2.2 IKK-regulating phosphatases ................................................................................. 30 1.2.3 Ubiquitin-mediated control of IKK ...................................................................... 32 1.3 FLICE-Like Inhibitory Proteins ................................................................40 1.3.1 Viral FLIPs ................................................................................................................ 40 1.3.2 Cellular FLIPs ........................................................................................................... 45 1.3.3 Regulation of cell death pathways by FLIPs ........................................................ 48 1.3.4 Regulation of autophagy pathways by FLIPs ....................................................... 55 1.3.5 Regulation of NF-κB pathways by FLIPs............................................................. 58 1.3.6 Regulation of MAPK pathways by FLIPs ............................................................ 59 1.3.7 FLIPs as promising targets for anti-cancer therapies.......................................... 60 1.4 HTLV-1 Tax: a functional analogue of the KSHV vFLIP .........................62 1.4.1 HTLV-1 biology ....................................................................................................... 62 1.4.2 HTLV-1 Tax ............................................................................................................. 62 1.5 Aims of the thesis .......................................................................................63

2.

MATERIALS AND METHODS ................................................. 64

2.1 Materials .....................................................................................................65 2.1.1 Molecular buffers and bacterial media .................................................................. 65 2.1.2 Antibodies ................................................................................................................. 66 2.1.3 Primers ....................................................................................................................... 67 2.1.4 Plasmids used in this study ..................................................................................... 69 2.2 Molecular biology....................................................................................... 71 2.2.1 Polymerase chain reaction (PCR) ........................................................................... 71 2.2.2 Site-directed mutagenesis ........................................................................................ 73 2.2.3 Restriction digestions ............................................................................................... 74 2.2.4 DNA ligations ........................................................................................................... 75 2.2.5 Annealing DNA oligonucleotides for subcloning into a plasmid ..................... 75 2.2.6 Agarose gel electrophoresis and recovery of DNA............................................. 75 6

2.2.7 Preparation, transformation and growth of competent bacteria ....................... 76 2.2.8 DNA purification and quantification .................................................................... 76 2.2.9 Extraction of cellular RNA and cDNA synthesis ............................................... 77 2.2.10 DNA sequencing ...................................................................................................... 77 2.3 Tissue culture .............................................................................................78 2.4 Lentivectors ................................................................................................78 2.4.1 Lentiviral transfer plasmids ..................................................................................... 78 2.4.2 Vector production .................................................................................................... 79 2.4.3 Viral titrations ........................................................................................................... 80 2.4.4 Cell transductions ..................................................................................................... 82 2.5 Generation of stable knock-down cell lines ...............................................83 2.5.1 pGIPZ ........................................................................................................................ 83 2.5.2 pHIV-SIREN ............................................................................................................ 83 2.5.3 shRNA sequences..................................................................................................... 84 2.6 Western blotting .........................................................................................86 2.7 Immunoprecipitation .................................................................................87 2.8 In vitro IKK kinase assay ...........................................................................88 2.8.1 Expression and purification of GST-IκBα(1-54) ................................................. 89 2.9 In vitro IκBα phosphorylation assay .......................................................... 91 2.9.1 Purification and expression of recombinant vFLIP, p22-FLIP and GB1p22FLIP ................................................................................................................................... 91 2.10 Immune complex dephosphorylation ........................................................92 2.11 Luciferase gene reporter assays .................................................................93 2.11.1 NF-κB reporter luciferase assays by transfection ................................................ 93 2.11.2 NF-κB reporter luciferase assays by transduction ............................................... 93 2.12 Statistical analysis ......................................................................................94

3. The Role of NEMO in IKK Activation by KSHV vFLIP and Cellular FLIPs ........................................................................................ 95 3.1 Introduction ...............................................................................................96 3.1.1 Molecular control of IKKs by NEMO ................................................................. 97 3.1.2 Interactions of the Tax, vFLIP and cFLIPs with NEMO ............................... 101 3.2 Aims of the chapter .................................................................................. 104 3.3 Results ...................................................................................................... 105 3.3.1 Generation of a NF-κB reporter luciferase assay system ................................. 105 3.3.2 Unlike vFLIP and Tax, cellular FLIPs require Ub-binding function of NEMO to activate NF-κB signalling ................................................................................................ 107 3.3.3 cFLIPs generate an active IKK without stable interaction with NEMO ...... 111 3.3.4 cFLIPL requires LUBAC to activate IKK .......................................................... 113 3.3.5 The UBAN domain of NEMO is dispensable for vFLIP activation of IKK 116 3.4 Discussion ................................................................................................ 120

4. The Role of Signalling Intermediates Upstream of IKK in NF-κB Signalling by KSHV vFLIP and Cellular FLIPs ................................. 125 4.1 Introduction ............................................................................................. 126 4.1.1 Caspase-8, FADD and RIP1 ................................................................................ 126 4.1.2 Atg3 .......................................................................................................................... 127 4.1.3 TAK1 and MEKK3 ............................................................................................... 127 4.2 Aims of the chapter .................................................................................. 127 7

Results ...................................................................................................... 128 4.3 4.3.1 cFLIPS and p22-FLIP activate IKK via a FADD-RIP1 complex.................. 128 4.3.2 Processing to p22-FLIP and p43-FLIP fragments is not necessary for NF-κB activation by cFLIPL ............................................................................................................. 133 4.3.3 Cellular FLIP variants require TAK1 to induce IKK ....................................... 136 4.4 Discussion ................................................................................................ 139

5.

Probing the Mechanism of IKK Activation of the KSHV vFLIP 143

5.1 Introduction ............................................................................................. 144 5.2 Aims of the chapter .................................................................................. 146 5.3 Results ...................................................................................................... 146 5.3.1 Recombinant vFLIP, but not p22-FLIP, can activate IKK when added to cell lysates................... ................................................................................................................... 146 5.3.2 Recombinant vFLIP can activate immuno-isolated IKK complexes in a cellfree assay system ................................................................................................................... 148 5.3.3 Phosphorylation of the IKKs at the activation loop is crucial for vFLIPinduced IKK activation ....................................................................................................... 150 5.3.4 Dimeric vFLIP-vFLIP interactions within the vFLIP-IKK signalosome are crucial for vFLIP activation of IKK .................................................................................. 153 5.3.5 A stapled peptide derived from HLX2 region of NEMO can efficiently block vFLIP-binding to IKK and its subsequent activation ..................................................... 159 5.4 Summary................................................................................................... 162

6. 6.1 6.2 6.3

7.

CONCLUSIONS AND FUTURE PERSPECTIVES............... 163 Insight into vFLIP activation of IKK ....................................................... 164 Insight into cFLIP activation of IKK ....................................................... 165 Therapeutic importance of our findings in viral and non-viral tumours 166

BIBLIOGRAPHAY .................................................................... 170

8

LIST OF FIGURES

Figure 1.1. Molecular architecture of the NF-κB, IκB and IKK family members ............. 19 Figure 1.2. A simplified scheme of the canonical and alternative NF-κB activation pathways. ........................................................................................................................................ 26 Figure 1.3. Viral and cellular FLIP proteins............................................................................. 43 Figure 1.4. Amino acid sequence alignment of the cellular and viral FLIPs....................... 44 Figure 1.5. Extrinsic and intrinsic cell death pathways. ......................................................... 50 Figure 1.6. FLIP proteins regulate the activity of the ripoptosome complex. .................... 54 Figure 1.7. Cellular and viral FLIPs block autophagy. ........................................................... 57 Figure 2.1. Lentiviral constructs. ............................................................................................... 82 Figure 2.2. Expression and purification of the GST-IκBα. ................................................... 90 Figure 3.1. Structure of NEMO and its complexes. ............................................................. 103 Figure 3.2. Generation of a cell-based NF-κB reporter luciferase assay. .......................... 106 Figure 3.3. Mutational studies on NEMO reveal differential NF-κB activation mechanisms for the KSHV vFLIP and cellular FLIP isoforms. .......................................... 109 Figure 3.4. NF-κB activation by A57L vFLIP and A56L cFLIP isoforms. ...................... 110 Figure 3.5. Cellular FLIPs constitutively activate IKK complex without stable association with NEMO. ................................................................................................................................ 112 Figure 3.6. Effects of knocking down LUBAC on the NF-κB activation ability of the Tax, vFLIP and cellular FLIPs. ................................................................................................. 114 Figure 3.7. Only cFLIPL is inhibited in CYLD-overexpressing HEK293 cells. .............. 115 ........................................................................................................................................................ 117 Figure 3.8. The UBAN domain of NEMO is dispensable for IKK activation by vFLIP and Tax. ........................................................................................................................................ 118 Figure 3.9. vFLIP binds to UBAN-deficient mutant of NEMO. ...................................... 119 Figure 3.10. Effect of the K192/195RR mutations on the NF-κB inducing ability of the cFLIP variants.............................................................................................................................. 124 Figure 4.1. FADD and RIP1 are required for NF-κB activation by cFLIPS and p22-FLIP. ........................................................................................................................................................ 130 Figure 4.2. A hydrophobic stretch of amino acids on the surface of DED2 is indispensable for NF-κB activation by cellular FLIP isoforms, but not KSHV vFLIP. .. 132

9

Figure 4.3. Unlike KSHV vFLIP, cellular FLIPs associate with a pre-assembled FADDRIP1 complex. ............................................................................................................................. 133 Figure 4.4. Non-cleavable mutant of cFLIPL activates NF-κB pathway in levels comparable to that of wild-type cFLIPL and p22-FLIP. ....................................................... 135 Figure 4.5. Cellular FLIPs require TAK1 for the activation of the NF-κB pathway. ..... 137 Figure 4.6. Analysis of vFLIP-, Tax- and TNFα-induced NF-κB activation in WT or TAK1 KD versions of the MEKK3+/+ and MEKK3−/− mouse embryonic fibroblasts. ... 138 Figure 5.1. Possible mechanisms underlying vFLIP-mediated activation of IKK........... 145 Figure 5.2. Recombinant KSHV vFLIP activates IKK when added to cell lysate. ......... 147 Figure 5.3. Direct in vitro activation of the IKK complex by recombinant vFLIP. ......... 149 Figure 5.4. Phosphorylation of the IKKs at their activation loop is required for vFLIPmediated activation of the IKK complex. ............................................................................... 152 Figure 5.5. Multimerisation of the vFLIP-NEMO complexes within the crystal............ 155 Figure 5.6. Interactions involved in the formation of higher order vFLIP-NEMO assemblies. .................................................................................................................................... 157 Figure 5.7. Mutations in the vFLIP-vFLIP and vFLIP-NEMO(2) interaction interfaces impair the vFLIP-induced activation of IKK. ........................................................................ 158 Figure 5.8. Crystal structure of the stapled NEMO peptide. .............................................. 160 Figure 5.9. The stapled NEMO peptide efficiently inhibits the vFLIP-induced IKK activation by blocking the interaction of vFLIP with NEMO. ............................................ 161 Figure 6.1. Our working model of IKK activation by the KSHV vFLIP and cellular FLIPs............................................................................................................................................. 168

10

LIST OF TABLES

Table 2.1. Buffers and bacterial media. ..................................................................................... 65 Table 2.2. Primary antibodies used for immunoblotting and immunoprecipitation. ........ 66 Table 2.3. HRP-conjugated secondary antibodies used for immunoblotting. .................... 67 Table 2.4. Primers used for amplifying cDNAs of genes of interest. .................................. 67 Table 2.5. Primers used for site-directed mutagenesis ........................................................... 68 Table 2.6. Sequencing primers ................................................................................................... 69 Table 2.7. Mammalian and bacterial expression vectors used in this project. .................... 69 Table 2.8. Phusion polymerase reaction mixtures .................................................................. 72 Table 2.9. goTaq polymerase reaction mixtures...................................................................... 72 Table 2.10. Cycling parameters for Phusion polymerase reactions ...................................... 73 Table 2.11. Cycling parameters for goTaq polymerase reactions .......................................... 73 Table 2.12. RT-PCR reaction mixtures (top) and thermocycler parameters (bottom). ..... 77 Table 2.13. Components of transfection mixture for LV production (amounts for a 15 cm plate) ......................................................................................................................................... 80 Table 2.14. Components of qPCR reaction for titration of lentivectors. ............................ 81 Table 2.15. shRNA-targeted sequences. ................................................................................... 84 Table 2.16. Western blot buffers and gel reagents .................................................................. 87 Table 2.17. Composition of kinase assay buffers. ................................................................... 88 Table 3.1. NEMO-interacting proteins................................................................................... 100

11

ABBREVIATIONS

aa aka Atg ATP BCA BSA CC CD cDNA cFLIP cIAP CMV cPPT CYLD DED DISC DMEM DNA dNTP DR DTT DUB E1 E2 E3 ECL EDTA EGTA ERK FBS FLICE FW GAPDH GFP GST HIV HLX HOIL-1L HOIP HRP

amino acid also known as autophagy-related protein adenosine 5'-triphosphate bicinchoninic acid bovine serum albumin coiled coil cluster of differentiation complementary DNA cellular FLICE-like inhibitory protein cellular inhibitor of apoptosis cytomegalovirus central polypurine tract cylindromatosis protein death effector domain death-inducing signalling complex Dulbecco's modified Eagle's medium deoxyribonucleic acid deoxynucleotide triphosphates death receptor dithiothreitol deubiquitinase ubiquitin-activating enzyme ubiquitin-conjugating enzyme ubiquitin ligase enhanced chemiluminescence ethylenediaminetetraacetic acid ethylene glycol-bis(-aminoethyl ether)-N,N,N’,N’-tetra acetic acid extracellular signal-regulated kinase foetal bovine serum Fas-associated death domain-like interleukin-1β-converting enzyme forward primer glyceraldehyde-3-phosphate dehydrogenase green fluorescent protein glutathione-S-transferase human immunodeficiency virus helical domain longer isoform of heme-oxidised IRP2 ubiquitin ligase-1 HOIL-1L interacting protein horseradish peroxidase 12

HSP HygoR IB Ig IKK-K IL-1 IP IκB JNK KA Kb KD kDa KO LB LPS LTR LUBAC LV LZ M MAPK MEF MEKK3 MOI mRNA mTOR NBD NEMO NF-κB NIK o C OD OE-PCR OIS PAGE PBS PCR PMA PMSF PP2A PuroR PVDF qPCR

heat shock protein hygromycin resistance gene immunoblotting immunoglobulin IKK-activating kinase interleukin-1 immunoprecipitation inhibitor of κB Jun N-terminal kinase kinase assay kilobases knockdown kilodalton knockout Luria Bertani lipopolysaccharide long terminal repeat linear ubiquitin chain assembly complex lentiviral vector leucine zipper molar mitogen-activated protein kinase mouse embryonic fibroblasts mitogen-activated protein/ERK kinase kinase 3 multiplicity of infection messenger RNA mammalian target of rapamycin NEMO binding domain NF-κB essential modulator nuclear factor κB NF-κB inducing kinase degrees Celsius optical density overlap extension-PCR oncogene-induced senescence polyacrylamide gel electrophoresis phosphate buffered saline polymerase chain reaction phorbol 12-myristate 13-acetate phenylmethanesulfonyl fluoride protein phosphatase type2A puromycin resistance gene polyvinylidene difluoride quantitative PCR 13

RIP RMPI RNA RPM RS RT RT SFFV SHARPIN shRNA SIN siRNA SMAC SP TAK1 TCR TEMED TGF-β TLR TNF TRAF TRAIL U UBAN UBD UV vFLIP VSV-G WT ZF

receptor-interacting protein Roswell Park Memorial Institute medium ribonucleic acid revolutions per minute reverse primer reverse transcriptase room temperature spleen focus-forming virus SHANK-associated RH domain interacting protein short hairpin RNA self-inactivating vector small interfering RNA second mitochondria-derived activator of caspases stapled peptide TGF-β-activated kinase 1 T cell receptor tetramethylethylenediamine transforming growth factor-β Toll-like receptor tumour necrosis factor TNF receptor-associated factor TNF-related apoptosis-inducing ligand unit ubiquitin-binding in ABIN and NEMO ubiquitin binding domain ultraviolet viral FLICE-like inhibitory protein vesicular stomatitis virus-glycoprotein wild-type zinc finger

14

CHAPTER

1 1. INTRODUCTION

15

1.1 NF-κB signalling pathway: an overview Nuclear Factor-κB (NF-κB) is one of the crucial signalling pathways which regulates a wide variety of physiological processes such as innate and adaptive immunity, proliferation, differentiation and cell death. In 1986, NF-κB was identified in David Baltimore’s laboratory as an inducible nuclear protein which binds to specific DNA sequences within κ light chain enhancer in B cells, referred to as κB sites (Sen and Baltimore, 1986a, 1986b). Extensive research over the past 27 years has revealed an enormous number of biological roles for this pathway in almost all cell types. The activation of NF-κB is strictly controlled at multiple levels by positive and negative regulatory elements whose malfunction can give rise to various pathologies, notably inflammatory diseases and cancers (Oeckinghaus et al., 2011). In the following sections, I provide a brief description of NF-κB signalling components and their mechanisms of action. 1.1.1 NF-κB related proteins In resting conditions, NF-κB is sequestered in cell cytoplasm by inhibitor of κB (IκB). Upon activation by various stimuli such as IL-1 or TNFα, the IκB is phosphorylated by IκB kinase (IKK) complex leading to its ubiquitination and subsequent proteasomal degradation. The released NF-κB is further regulated by post-translational modifications and migrates to nucleus where it promotes the transcription of target genes. In principle, three major families of protein form the backbone of NF-κB signalling network: NF-κB transcription factors, IκB family and IKK complex (Hayden and Ghosh, 2008).

1.1.1.1 NF-κB transcription factors In mammals, NF-κB family consists of five transcription factors: RelA (p65), RelB, c-Rel, NF-κB1 (p105/p50) and NF-κB2 (p100/p52) encoded by RELA, RELB, REL, NFKB1 and NFKB2 genes, respectively (Gilmore, 2006) (Figure 1.1A). Except for RelA, transcription of the other subunits is upregulated by NF-κB, producing a positive feedback loop on stimulated cells (Huxford et al., 2011). Synthesised as large precursor proteins, p105 and p100 undergo proteolytic processing to generate the transcriptionally active mature polypeptides, p50 and p52, respectively (Ghosh et al., 1998). Processing of

16

p105 occurs constitutively (Karin and Ben-Neriah, 2000; Lin et al., 1998), whereas p100 processing is mediated through regulated signals (Liou et al., 1994). All members of the NF-κB family share a conserved N-terminal region of roughly 300 amino acids, known as the rel homology domain (RHD) which is responsible for hetero- and homo-dimerisation, nuclear localisation, DNA binding and association with IκBs (Hayden and Ghosh, 2004). In RelA, RelB and c-Rel, the RHD is followed by a transcription activation domain (TAD) which mediates their association with trans-acting factors leading to positive regulation of gene expression (Hayden and Ghosh, 2008). On the contrary, p50 and p52 which lack a TAD fail to mediate transcription as homo-dimers and may act as repressors unless hetero-dimerised with one of the TAD containing NFκB subunits or other cofactor-recruiting proteins (Hayden and Ghosh, 2008; Zhong et al., 2002). Some studies, however, suggest that these homo-dimers are able to mediate activation through interaction with nuclear IκB proteins which can act as coactivators (Smale, 2012). Unlike the Rel proteins, p100 and p105 precursors contain an ankyrin repeat domain (ARD) in their carboxy terminal halves, similar to that found in the IκB family (Hoffmann et al., 2006). RelB is unique in that it contains a further N-terminal lucine zipper-like (LZ) motif which - in addition to the TAD - is necessary for its full activation (Hayden and Ghosh, 2008). Through combinatorial interactions, NF-κB subunits can form up to 15 distinct homo- and hetero-dimers (Gilmore, 2006). Although 12 out of the 15 possible dimers have been identified in various tissues, other yet undetected dimers may exit in some specific cellular conditions (Huxford et al., 2011). RelA: p50 is the most abundant of NFκB dimers, identified in almost all cell types (Oeckinghaus and Ghosh, 2009). The composition of NF-κB dimers can vary depending on cell type, stimulus and duration of signalling (Sen and Smale, 2009). In addition to the combinatorial specificity, various posttranslational modifications of NF-κB polypeptides as well as diverse interactions with other coactivators contribute further to remarkable complexity of cell- and stimuli-specific NF-κB responses (Sen and Smale, 2009).

1.1.1.2 IκB family IκBs consist of structurally related proteins, characterised by the presence of a conserved ARD that mediates their interaction with NF-κB subunits (Li et al., 2006a; Zheng et al., 2011). Based on their subcellular localisation as well as functional and

17

structural characteristics, the IκB family can be subdivided into two groups of typical and atypical IκBs (Figure 1.1B). IκBα, Iκβ and IκBε represent the typical cytoplasmic IκBs that function in part by masking the nuclear localisation sequence (NLS) of NF-κB polypeptides - contained within the RHD - and therefore, inhibit their nuclear translocation (Vallabhapurapu and Karin, 2009). In case of IκBα, the NLS masking is incomplete but a nuclear export sequence (NES) within IκBα mediates quick export of NF-κB:IκBα complexes from the nucleus. This leads to continuous shuttling of the complexes between cytoplasm and nucleus. Upon stimulation-induced degradation of IκBα, this balance shifts in favour of nuclear localisation and leads to transactivation (Hayden and Ghosh, 2008). Selective binding of the typical IκBs to specific homo- or hetero-dimers ties them to their distinct functions(Malek et al., 2001; Tran et al., 1997). Discrete degradation dynamics as well as different stimuli-dependant expression modes contribute further to specificity of their transcriptional activities (Hinz et al., 2012). The atypical IκBs consist of IκBζ, IκBη, IκBNS and B cell lymphoma-3 (Bcl-3) that all predominantly reside in the cell nucleus. Nuclear IκBs can bind to DNA-bound NFκB, inhibit their degradation and mediate their interaction with various cofactors resulting in positive or negative regulation of transcription (Ghosh and Hayden, 2008). Expression of atypical IκBs, except for IκBη, is limited in resting cells but can be greatly induced upon NF-κB stimulation (Hinz et al., 2012). Through their C-terminal ankyrin repeat motifs, NF-κB precursors p100 and p105 can also function as IκB-like proteins, sequestering their dimeric partners in the cytoplasm (Naumann et al., 1993a, 1993b) (Figure 1.1B). In contrast to typical IκBs which demonstrate subunit-specific inhibitory function, p105 binds to and inhibits all NF-κB subunits including its own processed form, p50. Recent studies have shown that multiple units of p100 and p105 can form large complexes (referred to as IκBsomes) that contain various NF-κB subunits. Different structural organisation and subunit binding preferences of atypical IκBs suggests that they might have distinct kinetic properties in activation and post-induction termination of NF-κB signalling as compared with typical IκBs (Huxford et al., 2011).

18

Figure 1.1. Molecular architecture of the NF-κB, IκB and IKK family members. The number of amino acids in each human protein is indicated on the right. A) All members of the NF-κB share a rel homology domain (RHD) that is required for their dimerisation, nuclear localisation, DNA binding and sequestration of IκBs. In RelA, RelB and c-Rel, this domain is followed by a transactivation domain (TAD) that mediates transcription initiation at κB sites-containing promoters. RelB additionally harbours a 19

leucine zipper domain in its N-terminus that is required for its maximal activity. The RHD in the p100 and p105 precursors is followed by a glycine-rich region (GRR), ankryin repeat motifs (ANK) and a C-terminal death domain (DD). The p52 and p50 subunits are shown below the p100 and p105 structures, generated after their proteolysis. B) IκB family members are characterised by the presence of multiple ANK motifs which mediate interactions with NF-κB dimers. IκBα and IκBβ contain an additional C-terminal PEST domain (sequences rich in proline (P), glutamate (E), serine (S), and threonine (T) residues) that is important for normal protein turnover. C) The core IKK complex consists of the enzymatic subunits, IKKα and IKKβ, and the regulatory subunit, NEMO. Serine residues within the activation loop of the IKKα/β kinase domain (KD) that undergo phosphorylation have been indicated. The previously designated leucine zipper (LZ) and helix-loop-helix (HLH) regions are shown in parentheses. ULD: ubiquitin-like domain, SDD: scaffold/dimerisation domain, NBD: NEMO-binding domain, HLX: helical domain, CC: coiled coli.

1.1.1.3 IκB kinase complex The IκB Kinase complex is the master regulator of the NF-κB signalling and consists of two enzymatic subunits: IKKα (IKK1) and IKKβ (IKK2) as well as a regulatory subunit known as NF-κB essential modulator (NEMO) or IKKγ (Scheidereit, 2006) (Figure 1.1C). Several biochemical studies on purified IKK complex have proposed a 2:1:1 ratio of the subunits where a dimer of NEMO associates with 1 IKKα heterodimerised with 1 IKKβ. However, other combinations of IKK core complex have also been suggested to exist, including the homodimers of IKKα and IKKβ either associated with or distinct from NEMO or separate (Liu et al., 2012). In vitro experiments provide evidence that IKK complexes can further assemble to form high-order structures through multimerisation (Drew et al., 2007). This can explain the high molecular weight of the endogenous IKK complexes (700-900 kDa) when separated by gel filtration chromatography. However, it is thought that the elongated shape of NEMO may also be the reason for the apparent molecular weight of IKK (Hayden and Ghosh, 2008). 1.1.1.3.1

The catalytic subunits

IKKα and IKKβ are structurally very similar and share 51% amino acid sequence homology (Mercurio et al., 1997). Both kinases contain an N-terminal kinase domain 20

(KD) -with two serine residues within the activation loop (S176 and S180 for IKKα, S177 and S181 for IKKβ) that require phosphorylation for the kinase activity, followed by a dimerisation domain and a C-terminal NEMO binding domain (NBD)(Israël, 2010; Liu et al., 2012; Scheidereit, 2006) (Figure 1.1C). IKKα, but not IKKβ, has a putative nuclear localisation signal (NLS) that has been linked to its NF-κB-independent nuclear activities (Sil et al., 2004). Crystal structures of IKKβ (from human and Xenopus laevis) were recently resolved, leading to a significant progress in understanding the domain organisation and function of this enzyme (Liu et al., 2013; Polley et al., 2013; Xu et al., 2011). Based on these studies, IKKβ has a trimodular architecture comprising the N-terminal KD, a central ubiquitinlike domain (ULD) and C-terminal elongated α-helical scaffold/dimerisation domain (SDD) (Figure 1.1C). Interestingly, neither of the previously predicted helix-loop-helix (HLH) and leucine zipper (LZ) motifs form these structures but they are part of the SDD (Xu et al., 2011). The ULD is required for catalytic activity of IKKβ and, together with SDD, is involved in determining the substrates specificity towards IκBα. The SDD mediates dimersiation of IKKβ which is required for kinase activation, but not important for maintaining the kinase activity once the activation loop is phosphorylated (Xu et al., 2011). In both IKKα and IKKβ, the NBD contains a shared six amino acid sequence (LDWSWL) that is necessary for their interaction with NEMO (May et al., 2002). Cell permeable peptides containing this sequence have been utilised to specifically disrupt IKK-NEMO interaction and prevent cytokine-induced NF-κB activation. Interestingly, competition assays using the peptide mimic of the NBD indicate a considerably weaker interaction of NEMO with IKKα compared to IKKβ (May, 2000). The higher affinity of IKKβ might be due to a unique 12 amino acid region that is not found in IKKα. This extension occurs immediately after the NBD and contains five negatively charged glutamic acid residues (Delhase, 1999). Remarkably, swapping the C-termini of IKKα and IKKβ generates an IKKα with IKKβ-like behaviour. These results suggest that differences in the NEMO binding affinity of IKKα and IKKβ might be responsible for their distinct functions (Kwak, 2000).

21

1.1.1.3.2

NEMO/IKKγ

The regulatory subunit of IKK complex, NEMO, is a 49 kDa polypeptide predicted to have two helical (HLX) domains interleaved with two coiled-coil (CC) domains, followed by a leucine zipper motif and a C-terminal zinc finger (ZF) (Zheng et al., 2011)(Figure 1.1C). Unlike IKKα and IKKβ, NEMO lacks any intrinsic enzymatic activity. The N-terminal region of NEMO (amino acids 40-120 contained within HLX1 region) mediates its binding to the C-termini of the catalytic subunits (May, 2000). The CC2 domain together with the adjacent LZ form the ubiquitin binding domain of NEMO, generally referred to as the NUB (NEMO ubiquitin binding), CoZi (coil-zipper domain) or UBAN (ubiquitin-binding in ABIN and NEMO)(Bloor et al., 2008; Ea et al., 2006; Wu et al., 2006a). The ZF motif was suggested by a recent study to be required for directing the substrate-specificity of IKKβ towards IκBα (Schröfelbauer et al., 2012). This domain has also been implicated in enhancing affinity of NEMO for binding to ubiquitin chains (Laplantine et al., 2009). Except for the C-terminus, the α-helical sequence of NEMO is capable of forming coiled-coil structure either alone or in association with a partner protein. The CC regions of NEMO are disrupted in multiple points by amino acids out of coiled-coil register. These breaks can serve as docking sites for interaction with various regulatory proteins and the flexible nature of CC region enhances the efficiency of these interactions (Ghosh et al., 2012). NEMO is encoded by the IKKBG gene located on the chromosome X. In humans, amorphic mutations of NEMO that lead to a lack of NEMO-dependant NF-κB activation are lethal in males but in females cause the disease incontinentia pigmenti (IP). IP is characterised by abnormalities of skin, hair, teeth, nails, and in some cases, neurological complications (Berlin et al., 2002). Hypomorphic mutations of NEMO that result in a weakened, but not obliterated, NF-κB induction have been associated with the X-linked recessive disease of anhidrotic ectodermal dysplasia associated with immunodeficiency (EDA-ID) in males. Symptoms of EDA-ID include severe defects in immunological functions, sparse hair, dental defects and hypohydrosis (Döffinger et al., 2001). Mutations in NEMO have been also identified in patients with Mendelian susceptibility to mycobacterial disease (MSMD) who have recurrent infections with bacteria of the tuberculosis family (Al-Muhsen and Casanova, 2008).

22

1.1.1.3.3

Other IKK-associated components

Apart from IKKα, IKKβ, and IKKγ which form the core of IKK complex, other proteins have been also reported to participate in the architecture and function of the complex. The chaperones heat shock protein 90 (HSP90) and HSP70 are two such components detected in association with the IKK complex. HSP90 constitutively associates with IKK through its co-chaperone Cdc-37 and has been suggested to stabilise the complex through facilitating its folding. Inhibition of HSP90 by Geldanamycin (an ATPase inhibitor of HSP90) suppresses NF-κB signalling in response to various stimuli such as IL-1, TNFα and PMA (Broemer et al., 2004; Chen et al., 2002). Unlike HSP90, HSP-70 appears to function as a NEMO interacting inhibitor of the IKK through preventing the formation of the complex (Ran et al., 2004). Another component that associates with IKK and regulates its function is a 105-kDa protein, ELKS (a protein rich in glutamate (E), leucine (L), lysine (K) and serine (S)).

Immunodepletion analyses

suggested that ELKS is a stoichiometric component of IKK (Häcker and Karin, 2006). siRNA-mediated knockdown of ELKS results in reduced levels of IKK activation in response to TNFα and IL-1. Recent studies support a role for ELKS in ATM- and NEMO-dependent NF-κB activation in response to genotoxic stress induced by DNA double stranded breaks (Hadian and Krappmann, 2011; Yang et al., 2011). Although the association of ELKS, HSP90 and HSP70 with IKK have been well-established, the detailed mechanisms by which they regulate IKK have yet to be identified. 1.1.2 Two pathways to NF-κB Numerous receptor-mediated cascades that lead to activation of NF-κB are classified into two major pathways, the canonical (or classical) and the non-canonical (or alternative) NF-κB pathway. These pathways are distinct in respect to the triggering stimuli, the IKK components involved and the targeted NF-κB subunits.

1.1.2.1 Canonical NF-κB pathway The canonical pathway is triggered by a broad range of stimuli including inflammatory cytokines (such as IL-1, TNFα, etc), pathogen-associated molecular patterns (PAMPs) and antigen receptors (Bonizzi and Karin, 2004).

This pathway is mainly

activated through phosphorylation of IκBs at serine residues (equivalent to Ser32 and Ser36 of IκBα) by IKKβ in a NEMO-dependant manner. The phospho-IκBs are then recognised and polyubiquitinated (at residues equivalent to Lys21 and Lys22 of IκBα) by 23

the E3 ubiquitin ligase complex SCFβTRCP (Skp1, Cdc53/Cullin1 and F-box protein β transducin repeat-containing protein). Subsequently, the K48-linked polyubiquitin chains mark IκB for degradation by 26S proteasome resulting in the release and nuclear translocation of RelA-containing NF-κB heterodimers, most commonly RelA:50 dimers (Vallabhapurapu and Karin, 2009) (Figure 1.2, left). Although IKKβ appears to be the predominant kinase of the canonical NF-κB pathway, several lines of evidence support a role for IKKα in this pathway. For example, receptor activator of NF-κB (RANK)-induced classical NF-κB activation in mammary cells depends on IKKα as the key kinase (Cao et al., 2001). Furthermore, IL-1 (but not TNFα)-induced canonical NF-κB induction has been shown to be intact in IKKβ deficient MEF cells (Solt et al., 2007). A recent study showed that IKKα is also involved in negative regulation of classical pathway through phosphorylating Tax1-binding protein 1 (TAX1BP1) and recruitment of the A20 deubiquitinase complex to IKK (Shembade et al., 2011). These findings indicate that IKKα may be involved in both activation and deactivation of canonical NF-κB signalling in response to at least a subset of stimuli. The ultimate outcome of the canonical pathway is transcriptional activation of the genes that are mainly involved in innate immunity such as pro-inflammatory cytokines (e.g., IL-1, IL-2, IL-6, TNF), chemokines (e.g.,CCL2, CCL3, CXCL8), leukocyte adhesion molecules (e.g., E-selectin, ICAM-1, and VCAM-1), and multiple pro-survival and antiapoptotic genes (e.g., Bcl-2, Bcl-XL, XIAP) (Bonizzi and Karin, 2004). Substantial increase in susceptibility to infections in conditional IKKβ or RelA KO mice reveals the importance of canonical pathway in immune functions (Pasparakis et al., 2006).

1.1.2.2 Alternative NF-κB pathway Unlike the canonical NF-κB pathway, the non-canonical NF-κB pathway is activated by a limited number of receptors that belong to the TNF receptor superfamily. These include CD40, lymphotoxin β receptor (LTβR), B-cell activating factor receptor (BAFFR), CD27, RANK, and Fn14 (Razani et al., 2011). Although most signals that activate canonical NF-κB pathway do not activate the non-canonical pathway, the noncanonical signals are able to activate both pathways (Bonizzi and Karin, 2004). The alternative NF-κB pathway strictly relies on IKKα and appears to be independent of IKKβ and NEMO (Senftleben et al., 2001) (Figure 1.2, right). Activation 24

of IKKα by NF-κB-inducing kinase (NIK), a member of the mitogen-associated protein 3 kinase (MAP3K) family, is an absolute requirement for switching on the non-canonical pathway (Xiao et al., 2001a, 2004). In resting cells, NIK is constitutively synthesised but cannot be detected due to a rapid degradation mediated by an E3 ubiquitin ligase complex composed of TNFα receptor associated factor-2 (TRAF2), TRAF3, cellular inhibitor of apoptosis-1 (cIAP1) and cIAP2 (Qing et al., 2005; Zarnegar et al., 2008). Evidences suggest that cIAP1/2 are responsible for the K48-linked ubiquitination of NIK while TRAF2 and TRAF3 cooperate to recruit NIK to cIAPs. Upon receptor ligation, the TRAF/cIAP complex is recruited to the receptor whereupon cIAP1/2 synthesises K48linked Ub chains on TRAF3 instead of NIK. The subsequent degradation of TRAF3 results in the accumulation of NIK that in turn phosphorylates and activates IKKα (Vallabhapurapu et al., 2008; Zarnegar et al., 2008). The activated IKKα then phosphorylates p100 (at ser866 and ser870) and marks it for proteasomal degradation, resulting in the release of the p52: RelB heterodimers that translocate to the nucleus and bind to the relevant DNA sequences (Senftleben et al., 2001; Xiao et al., 2004). The activated IKKα also phosphorylates NIK that leads to its destabilisation and downregulation of downstream signalling events (Razani et al., 2010). Biological roles of the alternative pathway that are mainly in regulating adoptive immunity include development of secondary lymphoid organs, B-cell maturation and survival, thymic epithelial cell differentiation, dendritic cells (DC) maturation and the osteoclastogenesis. Moreover, recent studies suggest a role or this pathway in regulating T-cell differentiation (Sun, 2012).

25

Figure 1.2. A simplified scheme of the canonical and alternative NF-κB activation pathways. Canonical pathway (right) is induced by a diverse range of surface receptors (e.g, TNFR or TLR) which converge on the tripartite IKK complex, comprising IKKα, IKKβ and NEMO. IKK activity results in phosphorylation of the IκB, followed by its K48-ubiquitination and proteasomal degradation. This leads to liberation of the Rel:p50 dimers which can then translocate to nucleus and induce plethora of target genes. Unlike the canonical cascade, the alternative pathway (right) is induced by a small subset of receptors and depends on the NIK-mediated phosphorylation and activation of IKKα homo-dimers. Activated IKKα phosphorylates p100 to induce its partial proteolysis to p52 subunit. Transcriptionally active p52 preferentially hetero-dimerises with RelB to migrate into the nucleus and initiate transcription of target genes. 26

1.2 Mechanisms behind positive and negative regulation of IKK To become active, the IKK complex requires phosphorylation of the T loop serines of at least one of the kinases. Similar to other kinases, this phosphorylation is more likely to confer activation through inducing conformational changes in the activation loop (Hayden and Ghosh, 2008). Mutation of these serines (S176 and S180 on IKKα, and S177 and S181 on IKKβ) to alanine prevents kinase activity of the IKK while replacement with phosphomimetic glutamates renders them constitutively active (Mercurio et al., 1997). The exact mechanisms by which the kinase subunits are phosphorylated remain controversial to date. However, three major activation mechanisms have been proposed: (i)

Direct phosphorylation of the kinase subunits by upstream IKK-activating kinases (IKK-K)

(ii)

Trans-autophosphorylation of IKK induced by oligomerisation

(iii)

Trans-autophosphorylation of IKK induced by conformational changes through protein-protein interaction or posttranslational modifications

These mechanisms are not mutually exclusive and might integrate in various celland pathway-specific manners (Häcker and Karin, 2006). Regulation of these mechanisms and the relevant responsible components including various kinases, phosphatases, adaptor proteins and polyubiquitin chains are the subject of the following discussions. 1.2.1 IKK-activating kinases The role of NIK for the phosphorylation and activation of IKKα in the alternative NF-κB pathway provided compelling evidence for the existence and function of IKK-Ks (Senftleben et al., 2001). Other studies have suggested a role for several other members of the MAP3K family as IKK-Ks in the canonical pathway. The most notable IKK-K is transforming growth factor-β kinase-1 (TAK1, aka MAP3K7) that was initially proposed to activate NF-κB through phosphorylating NIK (Ninomiya-Tsuji et al., 1999). Later on, in vitro experiments showed that TAK1, alongside its regulatory cofactors TAK1-binding protein1 (TAB1) and TAB2, can directly phosphorylate IKKβ on the activation loop in a ubiquitin-dependent manner (Wang et al., 2001). In cells, TAB2 or its homologues TAB3, but not TAB1, link TAK1 to upstream components by binding to K63-linked polyubiquitin chains on TRAF6 (in IL-1 signalling) or TRAF2 and RIP1 (in TNFα signalling) (Ea et al., 2006; Ishitani et al., 2003; Takaesu et al., 2000). This bridging is 27

mediated by a highly conserved C-terminal zinc finger domain on TAB2/3 (Kanayama et al., 2004). The mechanisms by which K63 ubiquitin chains induce TAK1 activation are not well-understood. However, it has been postulated that TAB2/3 binding to ubiquitin chains facilitates oligomerisation of TAK1 complexes leading to autophosphorylation and activation of TAK1 (Chen et al., 2006). In support of this hypothesis, TAK1 becomes phosphorylated at threonine187 within its activation loop and a mutation of this residue blocks its kinase activity (Singhirunnusorn et al., 2005). siRNA-mediated silencing of TAK1 as well as studies with deletion mutants of functional domains indicated that TAK1 is required for NF-κB activation (Besse et al., 2007; Takaesu et al., 2003). Furthermore, analysis of MEFs derived from TAK1-deficient mice supported a role for this enzyme in IKK activation by stimuli such as IL-1 and TNFα. Nevertheless, animal studies suggest that TAK1 might be dispensable in vivo, at least in some cell types. For instance, absence of TAK1 does not impair BCR-induced NF-κB activation (Sato et al., 2005). In addition, deletion of TAK1 blocks TCR induced NF-κB signalling in thymocytes, but not in effector T cells (Liu et al., 2006; Wan et al., 2006). Based on these results, it seems plausible that the requirement for TAK1 to activate IKK may be cell type-specific. Other studies have provided a role for mitogen-activated protein/ERK kinase kinase 3 (MEKK3, aka MAP3K3) as an IKK-K. Initial studies with MEKK3-/- fibroblasts suggested its importance in TNFα-induced NF-κB activation downstream of TRAF2 and RIP1. These cells showed reduced levels of IκBα degradation and the p65/p50 DNA binding in gel shift assays was severely impaired in response to TNFα. Since MEKK3 physically interacted with RIP1, it was proposed to mediate RIP1 signalling to IKK (Yang et al., 2001a). Later, Blonksa et al., were able to restore NF-κB activation in RIP1defiecient cells using a fusion protein composed of full length MEKK3 and RIP1 death domain (Blonska et al., 2005). This protein could directly associate with TRADD, indicating that RIP1 is likely responsible for recruiting MEKK3 to the TNFα receptor complex. MEKK3 is also implicated in IKK activation by IL-1R/TLR pathways and lysophosphatidic acid (LPA) (Huang et al., 2004b; Sun et al., 2009a). Similar to TAK1, MEKK3 interacts with TRAF6 upon induction of IL-1R/TLR pathways (Huang et al., 2004b). Yao et al., described independent pathways for IL-1 induced NF-κB activation by TAK1 and MEKK3. The TAK1-mediated pathway was shown to result in activation of 28

IKKβ, followed by IκBα phosphorylation and degradation; whereas the MEKK3dependent pathway led to NF-κB activation by NEMO phosphorylation and IKKα activation resulting in IκBα phosphorylation and dissociation from NF-κB subunits without its degradation (Yao et al., 2007). LPA-induced IKK activation, on the other hand, appears to be dependent on MEKK3, but not TAK1 (Sun et al., 2009a). Despite these findings, the physiologic importance of MEKK3 in NF-κB pathways has yet to be determined in vivo. Animal studies are partially hindered by the fact that MEKK3 knockout mice die at embryonic day 11 due to defects in angiogenesis and cardiovascular development (Yang et al., 2000). Two other MAP3Ks, MEKK1 and MEKK2, also exhibit ability to activate the IKK complex. Overexpression of MEKK1 and MEKK2 results in phosphorylation of IKKα/β in cells (Lee et al., 1997, 1998; Zhao, 1999). Although recombinant MEKK1 is shown to activate IKKs in vitro, there is no evidence for direct phosphorylation of IKKs by MEKK2. An analysis of biphasic cytokine-induced NF-κB activation demonstrated a mechanism in which MEKK2 regulates delayed NF-κB responses by assembling into IκBβ:NF-κB/IKK complexes; while MEKK3 inducibly associates with IκBα:NF-κB/IKK complexes and mediates rapid activation of NF-κB (Schmidt et al., 2003). Taken together, results from different studies suggest that IKK-Ks might have either redundant roles, cooperate to signal to IKK or function in distinct NF-κB pathways. Depending on the cell- and pathway-specific conditions, all three possibilities may exist. That being said, the exact role of IKK-Ks in IKK activation is still a matter for debate. Considering that IKK complexes can be activated by artificially enforced oligomerisation in vitro (Tang et al., 2003), IKK-Ks may contribute to amplification of IKK activity where it has been already initiated by trans-autophosphorylation. These functions, nevertheless, might be interchangeable in cells. Attempts to identify additional IKK-phosphorylating kinases have been hampered in part by the limitations of siRNA screenings. This is due to the fact that achieving complete protein knockdown in the whole cell population is almost impossible and the residual expression of a kinase is often sufficient to activate a signalling pathway normally (Liu et al., 2012).

29

1.2.2 IKK-regulating phosphatases Negative regulation of the MAPK signalling pathways by numerous MAPK phosphatases (MKPs) is well-established (Liu et al., 2007). By contrast, only a few phosphatases have been implicated in the modulation and termination of NF-κB pathway. The most prominent of all is a ubiquitously expressed serine/threonine phosphatase known as protein phosphatase type2A (PP2A). PP2A exists as heterotrimeric holoenzymes composed of one catalytic C subunit (PP2Acα or β isoforms) and a scaffolding A subunit (PR65α or β) which together form the core dimer, as well as a regulatory B subunit (Janssens and Goris, 2001). There are at least 18 regulatory B subunits whose binding to the ‘AC’ core is speculated to control PP2A substrate selectivity, catalytic activity and subcellular localisations (Cho and Xu, 2007). Several early studies associated PP2A activity with down-regulation of NF-κB on the observation that treatment with PP2A inhibitors enhanced IKK activation and blocked RelA phosphorylation in response to cytokines (DiDonato et al., 1997; Maggirwar, 1995; Yang et al., 2001b). In addition, recombinant PP2A could inhibit IKK activity and dephosphorylate RelA in vitro (DiDonato et al., 1997; Li et al., 2006b). In agreement with these findings, a more recent study demonstrated that siRNA-mediated silencing of various catalytic and regulatory PP2A subunits results in prolonged TNFαinduced NF-κB activation. Using co-immunoprecipitation and in vitro phosphatase assays, distinct PP2A complexes including PP2Acβ/PP2R1A , PP2Acα/PP2R1B

and

PP2Acα/PP2R1A/PP2R5C were detected to associate with and dephosphorylate IKKβ (on S181), RelA (on S536) and TRAF2 (on T117), respectively (Li et al., 2006b). In Tcells, through the regulatory subunit PP2R1A, PP2A interacts with Carma1 and removes the PKCθ-dependent phosphorylation of this protein on serine 645, thereby inhibiting the TCR-induced IKK activation (Eitelhuber et al., 2011). These results suggest that different combinations of PP2A holoenzymes may operate at different levels of the NF-κB pathway. Although dephosphorylation of IKKβ by PP2A has been reported by multiple research groups (Hong et al., 2007; Prajapati et al., 2004; Witt et al., 2009), some contradictory reports also exist. In a study by Sun et al., PP2A did not target IKKβ but rather dephosphorylated the upstream IKK-K, MEKK3. PP2A was shown to physically bind to phosphorylated MEKK3 and specifically dephosphorylate it on threonine516 and serine520 (Sun et al., 2010). Another study from the same group described the 30

magnesium-dependant phosphatases PPM1A and PPM1B, but not PP2A, as IKKβ phosphatases in the TNFα-induced NF-κB pathway (Sun et al., 2009b). Interestingly, PP2A is also implicated in positive regulation of the IKK activity (Kray et al., 2005). This is presumably mediated by reversing the inhibitory phosphorylation of NEMO and IKKβ following stimulus-dependant NF-κB activation (See 3.1.1.2). PP2A was detected in stable complexes with NEMO in resting conditions and the deletion of the putative PP2A-binding region on CC1 region of NEMO (amino acids 121-179) resulted in reduced IKKβ phosphorylation and activation in response to TNF-α (Kray et al., 2005). The discrepancies in the reported functions of PP2A and other IKK-regulating phosphatases may be explained by the distinct patterns of the phosphatase expression and/or substrate-specificity within different NF-κB pathways and cell types. PP1 is another phosphatase demonstrated to associate with and dephosphorylate IKKβ. The adaptor protein CUE domain-containing 2 (CUEDC2) mediates this interaction by recruiting GADD34, a regulatory subunit of PP1, to the IKK complex (Li et al., 2008a). Formation of the ternary complex of IKK/PP1/CUEDC2 was shown to retain IKK in an inactive, non-phosphorylated state but following TNFα stimulation, IKK was transiently dissociated from complex and bound to TRAF2, indicating that PP1 is likely responsible for controlling basal levels of IKK activity.. A recent siRNA screen conducted to identify NF-κB modulating phosphatases in T-cells pulled out PP4R1A as another negative regulator of IKK. PP4R1 which stably and specifically binds to the catalytic subunit PP4c was shown to associate with IKK complex subsequent to TCR-induced NF-κB activation and direct PP4c to dephosphorylate and deactivate IKK complex (Brechmann et al., 2012). Interestingly, PP4c has been linked to positive regulation of NF-κB based on the evidence that it can remove inhibitory phosphate groups from threonine435 of RelA (Yeh et al., 2004). Since PP4c subunit participates in a wide collection of PP4 holoenzymes, the associated regulatory subunits most likely determine the functional outcome of the phosphatase in NF-κB pathways (Chowdhury et al., 2008; Gingras et al., 2005; Lee et al., 2010). Wild-type p53-induced phosphatase1 (WIP1), which belongs to the magnesium dependent PP2C family of phosphatases, was found to directly interact with RelA and dephosphorylate it on serine536 in response to TNFα stimulation. WIP1 can also 31

dephosphorylate p38, which appears to be important for induction of a subset of NF-κB target genes such as IL-6, ICAM, and IRF1. Mice lacking WIP1 were reported to have hyperactivated immune responses based on the evidence that following LPS challenge, splenocytes of WIP1−/− mice produced higher levels of κB-dependent inflammatory cytokines, compared with WIP1+/− animals (Chew et al., 2009). However, in contrast to these results, Choi et al., showed that both T- and B-cells exhibit compromised functions in WIP1-deficient mice (Choi et al., 2002). A few other phosphatases such as Shp-2, PP6c and PTPN21, have been associated with negative regulation of the NF-κB pathway (Li et al., 2006b; Stefansson and Brautigan, 2006; You et al., 2001). However, more work is needed to elucidate distinct roles of these and other NF-κB-regulating phosphatases, in tissue- and pathway-specific conditions. 1.2.3 Ubiquitin-mediated control of IKK The ability to bind to ubiquitin chains, in free form or conjugated to other proteins, serves as a major mechanism in activating IKK. Ubiquitin-mediated control of IKK that in large depends on NEMO has been suggested to be involved in all possible IKK activation processes of oligomerisation, conformational change induction and recruiting IKK-Ks.

1.2.3.1 The ubiquitin system Ubiquitin (Ub) is a highly conserved 76 amino acid protein that is expressed in all eukaryotic cells (Hershko and Ciechanover, 1998). The term ‘Ubiquitination’ or ‘Ubiquitiylation’ refers to the covalent attachment of ubiquitin molecules to target proteins that is executed by a concerted action of three enzymes (Hoeller et al., 2006). First, the E1 ubiquitin-activating enzyme is loaded with ubiquitin through formation of a thio-ester bond between the C-terminal Glycine of the ubiquitin and the catalytic site Cysteine of E1. The activated ubiquitin is then discharged onto the active site of a ubiquitin-conjugating enzyme (E2), generating an E2-ubiquitin thiol-ester. Finally, an E3 ubiquitin ligase transfers the ubiquitin molecule to the target protein by forming an isopeptide bond between C-terminal carboxyl group of the ubiquitin and the ε-amino group of a lysine residue on a target protein (Ciechanover et al., 1982; Hershko, 1983; Hershko et al., 1983). Interestingly, some proteins without lysine residues have been also

32

found to be ubiquitinated. In these cases, serine or threonine residues are likely to participate in conjugation to ubiquitin (Cadwell and Coscoy, 2005; Wang et al., 2007b). To date, two E1s and approximately 50 E2s have been identified in mammals. E3s constitute the largest and the most diverse group of ubiquitin editing enzymes with over 600 members (Bhoj and Chen, 2009; Malynn and Ma, 2010). The E3 enzymes are classified into two families: the really interesting new gene (RING)-type and homologous to the E6-associated protein C terminus (HECT)-type E3 ligases (Bernassola et al., 2008; Petroski and Deshaies, 2005). The HECT-type ligases contain a conserved cysteine residue at the C-terminal part of the HECT motif that forms a thiol-ester bond with ubiquitin. HECT-domain bound ubiquitin can be transferred directly onto target protein (Pickart, 2001). Unlike HECT-type E3s, members of the RING-type family do not appear to form thiol-ester intermediates; instead they serve as scaffolds that bring the E2 and the target protein into close proximity. The RING finger domain contains a conserved pattern of cysteine and histidine residues whose folding allows coordination of two zinc cations (Deshaies and Joazeiro, 2009). Substrate proteins can be modified either by a single ubiquitin molecules (monoubiquitination) or ubiquitin polymers (polyubiquitination) (Haglund and Dikic, 2005). Each type of modification leads to a distinct regulatory fate. Monoubiquitination has been shown to regulate receptor endocytosis, vesicle sorting, gene silencing and DNA repair events [reviewed in (Hicke, 2001)] . It has been also implicated in regulating enzymatic activity of IKKβ (Carter et al., 2005) and transcriptional activity of the viral oncoprotein HTLV-1 Tax (Gatza and Marriott, 2006; Gatza et al., 2007). Ubiquitin harbours seven lysine residues (K6, K11, K27, K29, K33, K48 and K63) all of which can serve as ubiquitin acceptor sites, promoting formation of seven distinct polyubiquitin chains. In addition, ubiquitin linkages can be formed in a head-to-tail configuration through the N-terminal amino group, to produce so called ‘linear’ or ‘M1-linked’ ubiquitin chains (Komander, 2009). Depending on the linkage type, ubiquitin chains adopt distinct structural and functional characteristics (Pickart and Fushman, 2004). Based on the early studies on K48-linked ubiquitin chains, the ubiquitin system was thought to be merely in charge of proteasomal degradation, however the identification and study of other linkage types has revealed diverse biological roles of ubiquitin. All eight types of ubiquitin linkages have been detected in vivo and are suggested to regulate functions such as DNA repair (K63), protein interactions (K33), signalling pathway activation (K63, K27, K11, linear), 33

trafficking (K63) and the well-known proteasomal degradation (K48 , K11, K29) (Behrends and Harper, 2011). To recognise the signals encoded by ubiquitin moieties, cells have developed a series of modular motifs known as ubiquitin binding domains (UBD) that non-covalently bind to different forms of ubiquitin and transduce their signals into specific cellular pathways (Hicke et al., 2005). Using biochemical and bioinformatics approaches more than 150 types of UBDs have been identified to date. These domains are quite diverse in size (20-150 amino acids), structure and the functions of the UBD-containing proteins (Dikic et al., 2009). UBDs exist in E3s, deubiquitinases and binding adaptor proteins and contribute to their specificity. Most often, the binding affinity of the UBDs and ubiquitin are relatively low (about 10-500 µM) (Hicke et al., 2005). Nevertheless, specific mutations of UBDs, which disrupt ubiquitin binding, lead to impairment of protein function in vivo indicating the physiologic importance of these weak interactions (Ea et al., 2006; Kanayama et al., 2004; Wu et al., 2006a). Most UBDs show little discrimination between different linkages; although some prefer a certain type of polyubiquitin chains (Haglund and Dikic, 2005). For example, the NZF domain of TAB preferentially binds to K63-linked chains; this binding is essential for TAK1 activation (Kanayama et al., 2004; Komander et al., 2009). On the other hand, UBAN domain which is present in proteins such as NEMO, optineurin and A20-binding inhibitor of NF-κB (ABIN), shows high affinities towards linear polyubiquitin chains (Nagabhushana et al., 2011; Nanda et al., 2011; Rahighi et al., 2009). Ubiquitination processes are employed at multiple stages of both canonical and non-canonical NF-κB pathways. Indeed, cooperative functions of numerous ubiquitin editing enzymes, ubiquitin-binding proteins and different polyubiquitin chains appear to be essential for both negative and positive regulation of NF-κB pathway.

1.2.3.2 Roles of the linear and K63-linked PolyUb chains in IKK activation Much of our understanding of how polyubiquitin chains activate IKK comes from studying TNF and IL-1 receptor induced NF-κB signalling. In TNF signalling, following the trimerisation of the receptor by TNF, adapter protein (TRADD) (Hsu et al., 1995) and RIP1 (Hsu et al., 1996a) are recruited the to the death domain of TNF receptor (TNFR). TRADD then mediates recruitment of E3 ubiquitin ligase TRAF2 (and TRAF5)(Ermolaeva et al., 2008; Tsao et al., 2000) which in turn provides a platform for 34

binding of two more E3s, cIAP1 and cIAP2 (Mace et al., 2010; Vince et al., 2009). Next, cIAPs synthesise K63-linked ubiquitin chains on RIP1 (and cIAPs themselves)(Bertrand et al., 2008; Varfolomeev et al., 2008) that serve as a scaffold to bring IKK and TAK1 to close proximity via binding to UBD of their respective subunits, NEMO and TAB2 (or TAB3). K63-linked ubiquitin chains are especially important for TAK1-TAB recruitment and activation since the regulatory subunit TAB2 binds specifically to this type of linkages and not any other (Kulathu et al., 2009). Similarly, K63 polyubiquitin chains play an essential role in IL-1 and Toll-like receptor (TLR) induced NF-κB activation. Upon IL1R/TLR ligation, adapter protein MyD88 is recruited to the receptor complex which further associates with IRAK1 and IRAK4 via another adaptor protein TRAF-interacting protein with a forkhead-associated domain (TIFA) (Chen, 2012). IRAK1 induces K63-linked autoubiquitination of TRAF6 which depends on the E2 complex, Ubc13/Uev1A E2s (Deng et al., 2000; Lamothe et al., 2007). Activated TRAF6 also catalyses unanchored K63-ubiquitin chains which together with TRAF6-bound ubiquitin chains can recruit TAK1 and IKK, thereby facilitating phosphorylation of IKK by TAK1 (Wang et al., 2001). Initially, it was thought that UBAN domain of NEMO specifically binds to K63ubiquitinated components. However, several groups later showed that UBAN is able to discriminate linear and K63-linked Ub chains and remarkably, has 100-fold higher affinity for the first of these (Lo et al., 2009; Rahighi et al., 2009). In keeping with these studies, later a 600 kDa E3 ubiquitin ligase complex, so called LUBAC (linear ubiquitin binding assembly complex) was reported to mediate linear chains synthesis (Kirisako et al., 2006) and activate NF-κB independent of K63 ubiquitination (Tokunaga et al., 2009). LUBAC is composed of three subunits: HOIL-1L (longer isoform of heme-oxidised IRP2 ubiquitin ligase-1), SHARPIN (SHANK-associated RH domain interacting protein) and the catalytic subunit HOIP (HOIL-1L interacting protein) (Gerlach et al., 2011; Ikeda et al., 2011; Kirisako et al., 2006; Tokunaga et al., 2011). Generation of linear linkages is determined by LUBAC independent of the E2 involved. To date, LUBAC is the only known E3 capable of synthesising linear polyubiquitin chains (Iwai et al., 2014). An interesting recent study showed that most of linear ubiquitin chains formed in response to IL-1 are covalently conjugated to K63-linked ubiquitin oligomers. Furthermore, it was demonstrated that while HOIL-1L preferentially binds to linear chains, the catalytic subunit HOIP specifically interacts with K63-linked ubiquitin chains 35

as the preferred substrate (Emmerich et al., 2013). These findings indicate that the K63linked ubiquitin chains formed by upstream E3s like TRAFs and cIAPs might serve as a platform for recruitment of other regulators such as TAK1 and LUBAC. The linear ubiquitin oligomers synthesised by LUBAC can then associate strongly with NEMO and put IKK into a context where it can be phosphorylated by upstream kinases like TAK1, MEKK2 or MEKK3. Alternatively, ubiquitin binding of NEMO may promote IKK activation through oligomerisation

or

inducing

conformational

changes,

leading

to

trans-

autophosphorylation of the kinase subunits. The latter hypothesis is supported by the notion that the recognition of linear di-ubiquitin by UBAN of NEMO induces straightening of the coiled-coil region (Rahighi et al., 2009). This conformational alteration may then extend towards N-terminus and provide enzymatic subunits with optimal positioning for trans-phosphorylation. Interestingly, mutating K270 of murine NEMO (equivalent to K277 of human NEMO) can overcome the requirement for ubiquitin binding and renders the IKK complex constitutively active in the absence of any inflammatory stimuli (Bloor et al., 2008). It is possible that the K270A NEMO mimics the conformational changes induced by the ubiquitin binding to WT NEMO. Collectively, these findings propose a model in which NEMO keeps the catalytic subunits in an inactive conformation and ubiquitin binding induces a conformational change that removes this inhibition to promote activation of IKKs by upstream kinases or transautophosphorylation. Besides binding to ubiquitin chains, NEMO is also directly ubiquitinated by LUBAC at lysine residues K285 and K309 (Tokunaga et al., 2009). Reconstitution of NEMO-deficient cells with K285R/K309R double mutant NEMO did not rescue NF-κB activation in response or IL-1 or LUBAC overexpression. Since K309 is located within UBAN, its ubiquitination prevents binding to linear chains while in case of K285, theoretically conjugation and binding to linear polyubiquitin can coexist (Rahighi et al., 2009). However, because of the parallel nature of binding, the UBAN motif of NEMO is unlikely to recognize linear chains in cis and binds to ubiquitin chains of another NEMO molecule which may lead to oligomerisation-induced activation of IKK (Iwai and Tokunaga, 2009). The significance of ubiquitin chain binding and conjugation in canonical NF-κB pathways manifests both in human genetics and mouse models. Mutations in the UBAN 36

domain of NEMO (D311N, D311G) with deleterious effects on ubiquitin binding ability of NEMO have been associated with EDA-ID (Döffinger et al., 2001; Hubeau et al., 2011). In addition, cpdm (chronic proliferatory dermatitis in mice) mice, which have mutations in the Sharpin gene (Sharpin cpdm/cpdm), develop inflammatory disorders, severe skin lesions and defects in secondary lymphoid organs (Gerlach et al., 2011; Seymour et al., 2007). Cells derived from SHARPIN deficient as well as HOIL-1 knockout mice are sensitive to TNFα-induced apoptosis and have defects in NF-κB activation (Tokunaga et al., 2009, 2011).

1.2.3.3 DUBs and the negative regulation of IKK Deubiquitinases (DUBs) are a group of cysteine- or metallo-proteases that reverse the activity of E3s by cleaving ubiquitin moieties form target proteins (Harhaj and Dixit, 2011). Counter-regulation of ubiquitination processes by E3s and DUBs plays a crucial role in the regulation of NF-κB pathway and thereby, innate and adoptive immune responses (Sun, 2008). Approximately 100 DUBs are encoded in the human genome which are classified into five families based on the domain structure: ubiquitin-specific proteases (USPs), ubiquitin C-terminal hydrolases (UCHs), ovarian tumour proteases (OTUs), Machado–Joseph disease protein domain proteases (MJDs) and the JAB1/PAB1/MPN domain-containing metallo-enzymes (JAMMs) (Nijman et al., 2005). Similar to E3s, DUBs display intrinsic specificity towards different types of linkages. Recognition and recruitment of various linkages is mainly mediated by UBD of the DUBs but in some cases depends on the ubiquitin-binding adapter proteins or selectivity of the catalytic core (Komander, 2010; Komander and Barford, 2008; Komander et al., 2008). Several DUBs have been implicated in the negative regulation of IKK including CYLD, A20, cellular zinc finger anti-NF-κB (Cezanne), ubiquitin-specific protease 11(USP11), USP15 and USP21. CYLD was originally identified as a tumour suppressor since the mutation of the encoding cylindromatosis gene predisposes individuals for familial cylindromatosis, a genetic disorder characterised by benign tumours of skin (Bignell et al., 2000). DUB activity of this enzyme is mediated by its C-terminal USP domain (Komander et al., 2008). siRNAmediated knockdown of CYLD enhances NF-κB activation in response to inflammatory stimuli while overexpression of CYLD, but not the mutants lacking DUB activity, reduces NF-κB activation (Kovalenko et al., 2003; Trompouki et al., 2003). Several studies show that CYLD deficient mice are highly susceptible to chemically induced colitis as well as 37

colon and skin tumours (Reiley et al., 2006; Zhang et al., 2006). CYLD also plays an important role in T-cell development and activation, by regulating lymphocyte-specific protein tyrosin kinase (LCK), an important kinase in T cell receptor signalling (Reiley et al., 2006). CYLD has been demonstrated to target multiple components of the NF-κB pathway for deubiquitination such as NEMO (Kovalenko et al., 2003; Saito et al., 2004), TRAF2 (Brummelkamp et al., 2003), TRAF6 (Jin et al., 2008), RIP1 (Wright et al., 2007), TAK1(Reiley et al., 2007), LCK (Reiley et al., 2006) and the NF-κB subunit co-activator Bcl-3 (Massoumi et al., 2006). CYLD directly interacts with NEMO and TRAF2; however, in certain pathways, it requires ubiquitin-binding adaptor proteins to ensure target specificity. Optineurin and p62 have been suggested to link CYLD with RIP1 and TRAF6, respectively (Jin et al., 2008; Nagabhushana et al., 2011). In vitro experiments indicate that CYLD preferentially removes linear and K63linked ubiquitin chains compared to K48-linked oligomers (Komander et al., 2009). Results from two other studies show that CYLD cleaves K48-linked ubiquitin chains of some targets both in vitro (Stokes et al., 2006) and in vivo (Reiley et al., 2006). Taken together, it is possible that linkage-specificity of CYLD is determined in part by the target proteins. A20, also known as TNFα-induced protein 3 (TNFAIP3), is an NF-κB inducible protein that contains an N-terminal OTU domain and seven C-terminal zinc fingers (Krikos et al., 1992). A20-deficient mice die prematurely due to extensive inflammation of numerous organs including liver, kidneys, intestines and bone marrow (Lee et al., 2000). These mice are also highly susceptible to sublethal doses of LPS and TNF-α, indicating the substantial role of A20 in terminating inflammatory stimuli. In humans, polymorphisms of the A20 gene have been associated with multiple autoimmune diseases such as systemic lupus erythematous, Crohn’s disease, psoriasis and rheumatoid arthritis (Arsenescu et al., 2008; Fung et al., 2009; Musone et al., 2008; Nair et al., 2009; Thomson et al., 2007). Furthermore, inactivating mutations of A20 are found in a large number of human lymphomas (Honma et al., 2009; Kato et al., 2009). The unique feature of A20 is that it harbours both DUB and E3 activities which are mediated by the OTU and ZF4 domains, respectively (Wertz et al., 2004). In TNFR1 signalling, A20 first removes the K63-linked ubiquitin chains from RIP1, it then synthesises K48-linked ubiquitin polymers on RIP1 which marks it for proteasomal degradation (Wertz et al., 2004). An alternative mechanism has been described for A20 38

inhibition of the IL-1R/TLR pathways, whereby it blocks TRAF6 ubiquitination by disrupting its association with the E2 enzymes, Ubc13 and UbcH5C (Shembade et al., 2010). Similarly, A20 can inhibit TRAF2 and cIAPs by promoting disassembly of the E2:E3 complexes and triggering ubiquitin-mediated proteasomal degradation of the E2 enzymes. These functions depend on both ZF4 and OTU catalytic domains (Bosanac et al., 2010; Shembade et al., 2010; Wertz et al., 2004). Interestingly, two groups showed that overexpression of A20 mutant (C103A) lacking DUB activity blocks NF-κB activation as efficiently as WT-A20 (Evans et al., 2004; Li et al., 2008b). Consistent with these findings, a recent study demonstrated a non-catalytic mechanism of IKK inhibition by A20 (Skaug et al., 2011). Skaug et al., showed that linear and K63-linked ubiquitin chains on NEMO recruit A20 (via its ZF4 and ZF7) to the IKK complex which then can inhibit phosphorylation of IKK by TAK1, without reducing RIP1 ubiquitination. A20 can also bind directly to the N-terminal region of NEMO; this weak interaction is further stabilised by the interactions between A20 and polyubiquitin chains (Skaug et al., 2011). It is possible that A20 binding to NEMO blocks the IKK oligomersiation or reverses the conformational changes required for the activation of IKK. Taken together, these results suggest that A20 might utilise distinct mechanisms to terminate IKK activation in different NF-κB pathways. Although A20 and CYLD have many overlapping targets, no obvious functional redundancy exists between these two DUBs. This is presumably due to the distinct temporal order of their function during inflammatory responses (Sun, 2008). While CYLD blocks spontaneous activation of NF-κB, A20 is induced upon NF-κB activation and is crucial to terminate the pathway in a negative feedback loop. Phosphorylation of CYLD by IKK has been shown to be required for TRAF2 ubiquitination and activation of NF-κB and JNK pathways (Reiley et al., 2005). Therefore, phosphorylation-dependant inactivation of CYLD might provide a window for NF-κB activation before signalinduced termination by A20. Linkage type specificity is another key difference between CYLD and A20. In vitro studies suggest that A20 preferentially cleaves K48-linked polyubiquitin chains (Komander and Barford, 2008; Lin et al., 2008) while CYLD hydrolyses both linear and K63-linked oligomers (Komander et al., 2008, 2009). In cells, A20 might depend on ubiquitin-binding protein adaptors such as TAX1BP1 and the ABIN-1 to ensure specificity (Mauro et al., 2006; Shembade et al., 2007; De Valck et al., 1999).

39

In addition to CYLD and A20, other DUBs have been also described to play a role in negative regulation of the NF-κB pathway [reviewed in (Sun, 2008)]. For instance, Cezanne inhibits TNFα-induced IKK activation by promoting deubiquitination of the TNFR- associated RIP1. Similar to A20, Cezanne is rapidly induced in response to TNFα, although unlike A20, its DUB activity is essential for mediating the inhibitory functions (Enesa et al., 2008; Evans et al., 2003). Recently, Cezanne was reported to preferentially hydrolyse K11-linked ubiquitin oligomers (Bremm et al., 2010); how this function of Cezanne might regulate the NF-κB pathway still remains unknown. USP21 is another DUB which inhibits TNFα-induced IKK activity by also removing polyubiquitin chains from RIP1 (Xu et al., 2010). Clearly, IKK regulating DUBs share many overlapping targets; however the linkage type-specificity and the distinct temporal order of activity might account for their specific functions.

1.3 FLICE-Like Inhibitory Proteins The first members of the FLICE-like inhibitory protein (FLIP) family were identified as viral genome products (vFLIPs), following a bioinformatic search conducted to recognise death effector domain (DED)-containing proteins that function as apoptosis regulators (Thome et al., 1997). Soon after, a consensus sequence from the viral FLIPs was used to screen human expressed sequence tags and several human homologues were identified, collectively named as cellular FLIPs (cFLIPs)(Irmler et al., 1997). All members of the FLIP family harbour two tandem DEDs in their N-terminal region, similar to that found in caspase-8 and caspase-10 (Fig 1.3). As suggested by structural properties of FLIPs, the initial bioactivity reported was an ability to interact with FADD (Fasassociated protein with death domain) and inhibit apoptosis induced by multiple death receptors (Bertin et al., 1997; Hu et al., 1997a; Irmler et al., 1997; Thome et al., 1997). However, since their discovery, many other biological roles have been described for different FLIPs. In the following sections, I will review the structure and multi-faceted regulatory functions of the cellular and viral FLIPs with a particular emphasis on KSHV vFLIP. 1.3.1 Viral FLIPs Viral FLIPs are present in several γ-herpesviruses and the human molluscipoxvirus (Figuire 1.3A). The vFLIP-encoding γ-herpesviruses consist of bovine herpesvirus-4 40

(BHV-4), herpesvirus samiri (HSV), equine herpes virus-2 (EHV-2), rhesus monkey rhadinovirus (RRV) and the Kaposi’s sarcoma-associated herpesvirus (KSHV). The genome of molluscum contagiosum virus (MCV) encodes two distinct vFLIPs: MC159 and the closely related MC160 (Searles et al., 1999; Thome et al., 1997). Unlike cFLIPs, the amino terminal DEDs of different vFLIPs are not identical and contain variable amino acid sequences (Figure 1.4). Furthermore, none of the vFLIPs contain a caspase-8like domain; instead, the serial DEDs are extended by C-terminal tails of variable lengths. During the early stages of a viral infection, death receptor-induced cell death can effectively demolish the infected cells. Not surprisingly, viruses have evolved various strategies (e.g., expressing vFLIPs) to resist cell death and thereby, facilitate the viral propagation and persistence. The ability of vFLIPs to effectively inhibit apoptotic pathways not only results in persistent infection, but also transforms the host cells (Thome et al., 1997). KSHV vFLIP is a notable example of a viral oncoprotein that plays indispensable roles in tumourigensis of the associated virus.

1.3.1.1 Kaposi’s sarcoma-associated virus KSHV (also known as human herpesvirus-8 (HHV-8)) is the causative agent of Kaposi’s sarcoma (KS), a neoplasm of lymphatic endothelial cells. KS is the most common type of malignancy in HIV patients, although it occurs in other immunosuppressive conditions like organ transplant. KSHV is also associated with two Bcell lymphoproliferative diseases, primary effusion lymphoma (PEL) and multicentric Castleman’s disease (MCD) (Mesri et al., 2010). Similar to the life cycle of other herpesviruses, KSHV displays both lytic and latent modes of infection. Expression of lytic genes (such as vIL-6, vIRFs, vCCLs and vGPCR) have been documented to contribute to KSHV oncogenicity by initiating the host signalling cascades involved in secretion of cell growth factors. However, KSHV genes pivotal for the viral genomic persistence and cellular transformation are mainly found among those expressed during latent infection (Wen and Damania, 2010). Indeed, KSHV infection is predominantly latent in the associated malignancies. Three KSHV genes are abundantly expressed in the latent mode of infection: vCyclin, latency-associated nuclear antigen (LANA) and vFLIP. These genes are encoded in one tricistronic transcript driven by a single promoter and collectively provide cells with proangiogenic and inflammatory signals, anti-apoptotic abilities as well as enhanced proliferative and growth signals [reviewed in (Cesarman, 2014)].

41

1.3.1.2 KSHV vFLIP KSHV vFLIP (also known as K13) is a 188 amino acid protein encoded by ORF71 and shares 33% sequence homology with the cellular homologues. It is a potent and specific activator of the NF-κB pathway (Chaudhary et al., 1999; Matta and Chaudhary, 2004) but also interferes with several other pathways such as autophagy, MAPK and death inducing signalling cascades (See 1.3.3-1.3.6) . Expression of vFLIP induces transcription of up to 200 NF-κB target genes, most notably cFLIP which can further enhance evasion of death receptor induced apoptosis (Punj et al., 2009). Our group has previously shown that NF-κB-mediated secretion of growth factors enables vFLIP to protect cells from detachment-induced apoptosis, or anoikis (Efklidou et al., 2008). Accumulating evidence support a role for vFLIP in KSHV-associated tumourigenesis through constitutive activation of NF-κB (Ballon et al., 2011; Baud and Karin, 2009; Chugh et al., 2005; Sun et al., 2003). A recent study on transgenic vFLIP knock-in mice indicated that expression of vFLIP in B cells is sufficient to induce B-cell malignancies in vivo (Ballon et al., 2011) . Furthermore, vFLIP accelerated the process of lymphomagenesis in Myc-transgenic mice which overexpress Myc in lymphoid organs (Ahmad et al., 2010). In agreement with these findings, pharmacologic or genetic inhibition of vFLIP-induced NF-κB activation leads to apoptosis in PEL cells (Godfrey et al., 2005; Guasparri et al., 2004; Keller et al., 2000). Besides anti-apoptotic effects, vFLIPmediated NF-κB activation suppresses the lytic reactivation of the KSHV and thereby, further contributes to maintenance of latent infection (Ye et al., 2008).

42

Figure 1.3. Viral and cellular FLIP proteins. A) Viral FLIPs are encoded by several γherpesviruses and the human molluscipoxvirus (MCV). Unlike cFLIPs which share a common N-terminal sequence of 202aa, tandem DEDs of different vFLIPs are not identical and contain variable amino acid sequences. B) From 13 splice variants of the cFLAR gene, only three are translated to protein which consist of a long isoform (FLIPL) and two short variants (cFLIPS and cFLIPR). cFLIPL (55kD) is structurally similar to procaspase-8 (C) but its C-terminal caspase domain is enzymatically inactive due to lack of a crucial catalytic cysteine. The N-terminal DEDs of cFLIPS, but not cFLIPL or vFLIPs, 43

are followed by a unique 19 amino acid sequence which plays an important role in ubiquitination of cFLIPS (at the indicated K192 and K195 residues) and its subsequent degradation. Upon interaction with caspase-8, cFLIPL can further be cleaved at positions D196 or D376 to produce p22- and p43-FLIP fragments, respectively. p22-FLIP is also a cleavage product of FLIPS/R. C) Domain organisation of procaspase-8

Figure 1.4. Amino acid sequence alignment of the cellular and viral FLIPs. Amino acid sequences of the shown FLIP proteins were retrieved from NCBI protein database (http://www.ncbi.nlm.nih.gov/protein) and aligned using PRALINE multiple sequence alignment software. Residues are colour-coded for the rank of conservation. The accession numbers for the protein sequences used for alignment were: cFLIP BAB32551.1, KSHV vFLIP AAD46498.1, RRV vFLIP AAF60069.1, HSV vFLIP CAC84369.1, EHV vFLIP NP_042671.1, MCV vFLIP 159 NP_044110.1, MCV vFLIP160 NP_044111.1.

44

1.3.2 Cellular FLIPs In 1997, cellular FLIP proteins were independently identified by several groups and therefore, are recognised by other names such as MRIT (Han et al., 1997), Usurpin (Rasper et al., 1998), iFLICE (Hu et al., 1997b), FLAME1 (Srinivasula et al., 1997), CLARP (Inohara et al., 1997), CASH (Goltsev et al., 1997) and Casper (Shu et al., 1997).

1.3.2.1 cFLIP isoforms and cleavage products Cellular FLIP is encoded by the CFLAR gene (CASP8 and FADD-like apoptosis regulator), located on the human chromosome 2q33-34, close to caspase-8 and caspase-10 genes. Proximity of these genes suggests that cFLIP may have emerged by duplication of procaspase-8/10 genes (Han et al., 1997; Inohara et al., 1997; Rasper et al., 1998; Srinivasula et al., 1997). To date, up to 13 splice variants of human cFLIP have been identified at mRNA level, three of which are expressed as proteins: Long isoform (cFLIPL: 55kDa), short isoform (cFLIPS: 26kDa) and cFLIP Raji (cFLIPR, 24kDa) first isolated from human Burkitt’s lymphoma B-cell line, Raji (Golks et al., 2005; Irmler et al., 1997) (Figure 1.3B). Unlike humans, mice do not express FLIPS and contain only two isoforms of FLIPL and FLIPR (Ueffing et al., 2008). The CFLAR gene contains 14 exons and the initially transcribed precursor mRNA undergoes alternative splicing to generate transcripts for different cFLIP isoforms. Inclusion of exon7 which harbours a stop codon, results in translation of cFLIPS, while skipping this exon generates FLIPL-encoding mRNA. Translation of a small part of intron6 generates the shortest isoform, FLIPR (Djerbi et al., 2001). Recently, a single nucleotide polymorphism (SNP) of the 3’ splice consensus of intron6 was identified controlling expression of FLIPS or FLIPR. Interestingly, the splice-dead SNP which causes the expression of FLIPR is present at higher frequencies in transformed B cell lines and is also associated with an increased risk of follicular lymphoma in humans (Ueffing et al., 2009). All cFLIP isoforms share an identical N-terminal sequence of 202 amino acids, which includes two DEDs, but have different C-terminal ends. The overall structure of cFLIPS is similar to vFLIPs but it contains a unique 19 amino acid C-terminal tail that is responsible for its ubiquitination and proteasomal degradation (Poukkula et al., 2005). In cFLIPL, the tandem DEDs are followed by a caspase-like domain (CLD; composed of p20 and p12) that is similar to the catalytic domain of procaspase-8/10 (Figure 1.3C). 45

However, this domain is catalytically inactive due to substitution of several amino acids essential for the caspase activity, including a cysteine in the Gln-Ala-Cys-X-Gly motif and a histidine in the His-Gly motif (Budd et al., 2006). In addition to the natural isoforms, two N-terminal cleavage products of cellular FLIPs have been detected in cells. Caspase-mediated cleavage of the cFLIP at positions D376 (on FLIPL) and D196 (on all isoforms), generates the proteolytic fragments p43FLIP (43kDa) and p22-FLIP (22kDa), respectively (Figure 1.3B) (Golks et al., 2006; Kataoka and Tschopp, 2004; Scaffidi et al., 1999).

1.3.2.2 Transcriptional and translational control of cFLIPs Expression of cFLIP proteins is tightly regulated in normal cells. This is achieved through numerous regulatory mechanisms at the transcriptional, translational and posttranslational levels (Safa et al., 2008). Several transcription factors are known to modulate the transcriptional activity of the CFLAR gene. NF-κB, CREB, FoxO, p63, p53, EGR1, NFAT, hnRNP K, AR and sp1 are among those which induce cFLIP transcription, while c-myc, Foxo3a, c-Fos, IRF5 and sp3 repress it (Shirley and Micheau, 2010). NF-κB activation is one the main inducers of cFLIPs; these proteins, in turn, elicit the NF-κB pathway, generating a positive feedback loop on stimulated cells (Micheau et al., 2001). Little is known about isoform-specific regulation of cFLIP transcription, although some scattered reports exist. For instance, the AP-1 complex has been shown to repress cFLIPL expression (Li et al., 2007), while E21F, a transcription factor involved in control of cell cycle, inhibits the expression of cFLIPS (Salon et al., 2006). Furthermore, an siRNA screening aimed at characterising the targets of p63 suggested that this protein upregulates cFLIPR transcription, while suppressing that of cFLIPS without altering the transcription levels of cFLIPL (Borrelli et al., 2009). The short isoform is shown to be regulated at translational level as well. Panner et al., demonstrated that in glioblastoma multiforme (GBM) cells, activation of the Akt/mTOR/S6K1 pathway results in polyribosomal accumulation of cFLIPS mRNA and therefore, increased expression of FLIPS. Inhibition of mTOR or its target S6K1 supressed translation of cFLIPS, but not cFLIPL, and led to TRAIL-sensitisation of GBM cells (Panner et al., 2005). An mTOR-independent pathway has been also described to regulate cFLIPS translation. In this pathway, activation of Ral and its effector protein

46

RalBP1, inhibits cdc42-mediated activation of S6K1 and thereby, down-regulates the expression of FLIPS (Panner et al., 2006).

1.3.2.3 Post-translational regulation of cFLIPs Cellular FLIPs exhibit a relatively short half-life. This has been indicated by rapid depletion of these proteins following treatment of cells with the inhibitor of protein synthesis cycloheximide (Kreuz and Siegmund, 2001; Micheau et al., 2001). Vice versa, cFLIPs are found to accumulate quickly in response to proteasome inhibitors such as MG-132, lactacystin, epoxomicin and bortezomib/Velcade® (Chang et al., 2006; Fukazawa et al., 2001; Kim et al., 2002; Perez et al., 2003). Post-translational modifications, most notably ubiquitination and phosphorylation, play crucial roles in regulating the turnover rate of different cFLIPs. cFLIP proteins are predominantly degraded via the ubiquitin-proteasome degradation system. cFLIPS is particularly more prone to ubiquitination and displays a considerably shorter half-life compared with cFLIPL (Poukkula et al., 2005; Schmitz et al., 2004). This is largely due to the unique 19 amino acid C-terminal sequence of cFLIPS that facilitates its ubiquitination at lysine residues 192 and 195, marking this protein for proteasomal degradation (Poukkula et al., 2005); the stability of cFLIPR is similar to cFLIPS (Golks et al., 2005; Ueffing et al., 2008). Both proteins have been shown to be targeted by the E3 ubiquitin ligase c-Cbl (Kundu et al., 2009; Zhao et al., 2013), although the E3 responsible for the ubiquitination of K192/195 residues remains to be identified. TNFα-mediated JNK1 (Jun N-terminal Kinase1) activation can enhance cFLIPL degradation by inducing the E3 ubiquitin ligase Itch (Chang et al., 2006). The caspase-like domain of FLIPL is necessary for interaction with Itch, highlighting once more the importance of C-terminal sequences of cFLIP isoforms in determining their stability. Phosphorylation events are also central to regulation of cFLIP levels. PKC phosphorylates cFLIP proteins at serine193 (Kaunisto et al., 2009). This results in decreased ubiquitination of all isoforms, although it only prolongs the half-lives of cFLIPS and cFLIPR. Since site-specific mutation of S193 does not affect recruitment of different isoforms to the DISC (death inducing signalling complex), PKC-mediated cFLIP phosphorylation may regulate apoptotic pathways only via rapid changes in cFLIPS levels (Kaunisto et al., 2009). This seems to be different from calcium/calmodulin-dependent kinaseII (CaMKII)-mediated phosphorylations of cFLIP proteins (at a residue other than 47

S193) which block DISC recruitment of both short and long isoforms (Yang et al., 2003). Recently, serine273 of cFLIPL was shown to be targeted by PI3K/Akt. Phosphorylation of

cFLIPL-S273

following

TNFα-induced

macrophage

activation

resulted

in

ubiquitination and proteasomal degradation of cFLIPL (Shi et al., 2009). This cFLIPL ubiquitinaton, however, was not dependent on JNK/Itch. Phosphorylation of cFLIPS is found to be crucial for mycobacterium tuberculosis-induced cell death in murine macrophages. M.tb-mediated TNF activation induces p38 and Abl, which in turn phosphorylate cFLIPS on Ser4 and Tyr211, respectively. This facilitates recognition of FLIPS by the E3 ubiquitin ligase c-Cbl, leading to its degradation and initiation of apoptosis (Kundu et al., 2009). Taken together, these studies elucidate how the cross-talk between phosphorylation and ubiquitination events results in quick isoform-specific changes in cFLIP levels. This provides cells with a crucial ability to generate rapid responses to cellular distress. 1.3.3 Regulation of cell death pathways by FLIPs Controlling the delicate balance between cell death and survival pathways is essential for development and homeostasis of multi-cellular organisms. To achieve this, cells have evolved numerous regulatory proteins such as FLIP, which can promote or inhibit demolition of cells in a certain set of conditions. Cell death can occur through several distinct routes (e.g, apoptosis, necrosis and autophagy-related cell death) and remarkably, FLIP proteins have been documented to be involved in regulation of all these types of cell demise.

1.3.3.1 Extrinsic and intrinsic apoptotic pathways Apoptosis is a programmed mode of cell death that enables multicellular organisms to dispose of unwanted cells with minimum damage to neighbouring cells (Taylor et al., 2008). Precise regulation of apoptosis is central to the development, differentiation and homeostasis of tissues. Apoptotic signalling pathways are classified into two distinct types: intrinsic pathways induced by factors such as DNA damage, UV or γ-radiation, chemotherapeutic drugs and cytokine deprivation (Figure 1.5, left), and extrinsic pathways initiated by death receptors (DR) (Krammer et al., 2007; Lavrik et al., 2005) (Figure 1.5, right). Proteins of the DR family belong to the TNFR superfamily. To date, eight members of this family have been described: TNFR1 (also known as DR1, p55, p60 and 48

CD120a), CD95 (also known as DR2, Fas and APO-1), DR3 (also known as APO-3, LARD, TRAMP and WSL1), TNF-related apoptosis inducing ligand receptor1 (TRAILR1; also known as DR4 and APO-2), TRAILR2 (also known as DR5, KILLER and TRICK2), DR6, ectodysplasin A receptor (EDAR) and nerve growth factor receptor (NGFR) (Lavrik et al., 2005). DR family are characterised by the presence of a C-terminal sequence of 80-100 amino acids, called the death domain (DD) that is responsible for transducing the apoptotic signals. Indeed, homotypic DD interactions connect DRs to adaptor proteins FADD and TRADD (Guicciardi and Gores, 2009). Upon ligation of CD95, TRAILR1 or TRAILR2, a multiprotein complex is formed near the plasma membrane known as death-inducing signalling complex (DISC). This complex is composed of DR, FADD, cFLIP and precursors of two initiator caspases: procaspase-8 and procaspase-10. While FADD and DR interact via DDs, cFLIP or procaspase-8/10 are recruited to FADD via the DEDs of each protein. Activation of the initiator caspases following DISC formation, results in the induction of effector caspases (caspase-3, -6 and -7), which in turn proteolyse a wide array of substrates leading to dismantling and packaging of cells into apoptotic bodies (Figure 1.5, right). In the intrinsic pathway, also known as the mitochondrial pathway, death signals are initiated by mitochondrial membrane permeabilisation (MMP) (Martinou and Green, 2001). Anti-apoptotic members of the Bcl-2 family (such as Bcl-2, Bcl-XL, BCL-W and MCL1), preserve the integrity of outer mitochondrial membrane by inhibiting the oligomerisation of proapoptotic Bcl-2 members, BAX and BAK. Upon arrival of a death signal, antiapoptotic Bcl-2 proteins are inhibited by Bcl-2 homology3(BH3)-only proteins (eg., BID, BAD, BIM and BMF) (Kuwana et al., 2005; Letai et al., 2002). This leads to formation of BAX-BAK oligomers within the outer membrane that allows for efflux of intermembrane space proteins such as cytochrome C, SMAC/DIABLO and AIF (apoptosis-inducing factor). Once released form mitochondria, cytochrome C mediates assembly of a caspase activation complex known as the apoptosome. This complex consists of about seven molecules of apoptotic protease-activating protein-1 (APAF1) and the same number of initiator procapspase-9. Similar to events followed by DISC assembly, activation of caspase-9 within the apoptosome induces a cascade of effector caspases that culminates in cell death (Taylor et al., 2008) (Figure 1.5, left). In some conditions, extrinsic death signals can activate the intrinsic pathway through caspase-8 mediated cleavage of the BH3-only family member, BID (BH-3 interacting domain death agonist). Truncated BID (tBID) can then promote release of intermembrane space proteins resulting in apoptosome formation and cellular demolition (Korsmeyer et al., 2000). 49

Figure 1.5. Extrinsic and intrinsic cell death pathways. Extrinsic apoptotic pathway (right) commences by engagement of death receptors (DRs) with their cognate ligands which then leads to recruitment of adaptor proteins (e.g, FADD and TRADD) and formation of DISC, consisting of DR, FADD, procaspase-8/10 and cFLIP isoforms. Autoprocessing and activation of the initiator procaspase-8/10, promotes the activation of downstream caspases, such as caspase-3, 6 and 7, for the execution of apoptosis. The extrinsic pathway is mainly under the control of cFLIP family which fine-tune the activity of initiator caspases at the DISC. In the intrinsic pathway (left) various stimuli such as UV or chemotherapeutic agents which provoke cell stress and damage activate one or more members of the BH3-only proteins. This in turn induces the assembly of BAK-BAX oligomers within mitochondrial outer membranes, leading to release of intermembrane proteins such as cytochrome C which promotes the assembly of apoptosome. Activation of caspase-9 within this complex then propagates a cascade of caspase activation, similar to final steps of the extrinsic pathway. Caspase-8-mediated cleavage of BID to generate tBID can link the extrinsic pathway to the mitochondrial cell death cascade; cFLIP can also inhibit this step.

50

Deregulation of cell death pathways resulting from insufficient or excessive apoptosis form the basis of multiple human disease such as neurodegenerative disorders, autoimmunity and cancer (Green and Evan, 2002; Vaux and Flavell, 2000). Therefore, apoptotic pathways are tightly controlled at multiple levels. For instance, members of the cIAP family, including NIAP, XIAP and cIAP1/2 have been shown to directly inhibit caspase-3,-7 and -9 (Liston et al., 1996). As mentioned above, proteins of the Bcl-2 family modulate the mitochondrial death programs. At the DISC level, apoptotic signals are regulated by FLIP proteins that facilitate or inhibit the activation of casapase-8. Nonetheless, DISC formation does not always result in cell death but it can be also directed to switch on several survival and proliferative pathways like NF-κB or MAPK. Remarkably, a collective body of research shows that FLIPs also play crucial roles in regulation of the DR-induced non-apoptotic pathways (Oztürk et al., 2012).

1.3.3.2 Roles of FLIPs in apoptosis Procaspase-8 activation, which is pivotal for DR-induced apoptosis, occurs through an “induced proximity” mechanism where DR-mediated homo-dimerisation of procaspase-8 leads to its self-cleavage and release of fully active catalytic subunits (p10 and p18) (Chang et al., 2003). Similar to procaspase-8, FLIPs are recruited to DISCassociated FADD via homotypic DED interactions. Hence, the anti-apoptotic function of FLIP proteins relies on their ability to restrain FADD-mediated homo-dimerisation and activation of procaspase-8 molecules (Krueger et al., 2001b). Although viral FLIPs, cFLIPR and cFLIPS have been well-established as potent inhibitors of procaspase-8 activation (Golks et al., 2005; Krueger et al., 2001a; Thome et al., 1997), function of cFLIPL has been inconsistently reported to be either pro-apoptotic (Goltsev et al., 1997; Han et al., 1997; Inohara et al., 1997; Shu et al., 1997) or antiapoptotic (Irmler et al., 1997; Rasper et al., 1998; Srinivasula et al., 1997). The proapoptotic role of cFLIPL is supported by the phenotype of cFLIP null mice which resembles that of mice lacking caspase-8 or FADD. These mice die at embryonic day 10.5 with impaired heart, vascular and haematopoietic development, suggesting a function for cFLIPL that is similar to FADD or caspase-8 (Yeh et al., 2000). Furthermore, expression of cFLIPL at physiological levels is suggested to mediate apoptosis by formation of catalytically active cFLIPL/caspase-8 heterodimers (Chang et al., 2002; Micheau et al., 2002). Nevertheless, the pro-apoptotic role of cFLIPL disagrees with the observation that cFLIPL-specific knockdown cells or MEFs lacking cFLIP are highly susceptible to DR51

induced cell death compared with WT cells (Sharp et al., 2005; Yeh et al., 2000). In addition, ectopic expression of cFLIPL at high levels is found to inhibit procaspase-8 activation via competing for FADD binding (Scaffidi et al., 1999). Despite these contradictory reports, several studies have now elucidated that depending on its expression levels cFLIPL can either promote or inhibit apoptosis (Chang et al., 2002; Fricker et al., 2010; Neumann et al., 2010). A recent study by Fricker et al., analysed the relationship between amounts of cFLIPL and the regulation of CD95induced cell death, using mathematical modelling combined with quantitative western blotting. The authors demonstrated that cFLIPL blocks apoptosis when highly overexpressed, whereas its moderate expression can promote cell death when accompanied with strong stimulation of CD95 or in the presence of high levels of cFLIPS/R (Fricker et al., 2010). Therefore, the cFLIPL role at the DISC is not only dosedependent but also may rely on other factors such as signal strength and the levels of other cFLIP isoforms. Biochemical assays as well as crystal structure studies on vFLIP MC159 propose that viral and cellular FLIPs may use different mechanisms to inhibit caspase-8 activation. The homotypic interactions between FADD-DED and procaspase-8-DED2 or cFLIPDED2 are mediated through a highly conserved hydrophobic patch (Berglund et al., 2000; Carrington et al., 2006; Eberstadt et al., 1998). In pull-down experiments, cFLIP competes away procaspase-8 in binding to FADD. In contrast, MC159 cannot compete procaspase8 away and binds to FADD-DED through an extensive area outside the hydrophobic patch interface (Yang et al., 2005). Therefore, MC159 (and perhaps other vFLIPs) may disrupt DISC assembly by preventing FADD-DED self-association rather than inhibiting procapase-8 binding (Yu and Shi, 2008). Overexpression of isolated FADD-DED or tandem DEDs of procaspase-8 results in self-assembly to produce long cytoplasmic filaments (also known as death effector filaments) which can induce apoptosis independent of death receptors (Siegel et al., 1998). Unlike these proapoptotic DEDs, which are found highly aggregated in vitro, isolated tandem DEDs from MC159 or EHV-2 vFLIP E8 do not form filaments in cell culture and appear as monomers in vitro (Li et al., 2006a; Yang et al., 2005). Indeed, these viral DEDs can effectively block self-assembly of DEDs from FADD or procaspase-8 to form death filaments (Siegel et al., 1998). Whether other viral FLIPs use the same DISC-inhibition strategy still remains unanswered.

52

1.3.3.3 Roles of FLIPs in necroptosis Regulated necrosis, referred to as necroptosis, is an alternative form of cell death that occurs when caspase-8 activity is inhibited and it depends on kinase activity of RIP1 and RIP3 (Cho et al., 2009; He et al., 2009; Zhang et al., 2009a). The embryonic lethality observed in FADD −/− and CASP-8−/− mice can be rescued by additional deletion of the kinases RIP1 or RIP3, indicating that FADD and caspase-8 inhibit the RIP1/3-mediated cell death during embryonic development (Kaiser et al., 2011; Oberst et al., 2011; Zhang et al., 2011). Although necroptosis has evolved as an immune defence mechanism against intracellular infections, it has been also implicated in causation of several pathologies such as ischemia–reperfusion injury, neurological and myocardial disease (Linkermann and Green, 2014). Necroptotic signalling cascades can be initiated by stimulation of DRs, TLRs, genotoxic drugs and some viral and bacterial pathogens (e.g. human simplex virustype1 and vaccinia virus) (Vanlangenakker et al., 2012). Similar to their role in apoptosis, cFLIP proteins can regulate necroptosis in an isoform-specific manner. Evidence for modulation of necroptosis by FLIP proteins (and a connection between apoptosis and necroptosis) comes from the detection of a cytoplasmic complex, termed the ripoptosome, induced by treatments of human tumour cells that deplete or inhibit cIAP proteins (Feoktistova et al., 2011; Tenev et al., 2011a). This depletion leads to the formation of a RIP1/FADD/procaspase-8 complex. cIAPs inhibit formation of this complex by ubiquitinating RIP1 that leads to ubiquitin-mediated recruitment of several kinases required for NF-κB activation (Bertrand et al., 2008; Varfolomeev et al., 2008). CYLD-mediated deubiquitination of RIP1, on the other hand, enables the kinase activity of RIP1 and necrosome assembly (Wang et al., 2008). In the absence of FLIP proteins, procaspase-8 molecules homodimerise in ripoptosome, promoting cell death by apoptosis. Recruitment of cFLIPL to this complex leads to partial caspase-8 activation, not enough to induce apoptosis, but sufficient to cleave RIP kinases and disassemble the ripoptosome, resulting in cell survival (Boatright et al., 2004; Oberst et al., 2011; Pop et al., 2011). cFLIPS, on the other hand, blocks caspase-8 activation and facilitates cell death through necroptosis (Feoktistova et al., 2011) (Figure 1.6). Similarly, vFLIP MC159 shifts the outcome of ripoptosome toward necroptosis (Feoktistova et al., 2012). Roles of other viral FLIPs, however, have yet to be examined.

53

Figure 1.6. FLIP proteins regulate the activity of the ripoptosome complex. Under baseline conditions, a small proportion of RIP1 is persistently modified to active conformation. This modified RIP1, also called open conformation, is targeted for Ubmediated degradation by cIAPs. Upon depletion of cIAPs by genotoxic stress, accumulation of modified RIP1 promotes formation of a ripoptosome signalling complex which consists of RIP1, FADD, caspase-8/10 and cFLIP isoforms. Different compositions of the caspase-8 and cFLIP isoforms bring about different levels of ripoptosome-associated caspase-8 activity and therefore, lead to different cell death responses. Procaspase-8 dimerisation within the complex forms active caspase-8 and results in cell death via apoptosis. However, cFLIPL within the ripoptosome decreases the caspase-8 activity to levels which are not sufficient for its apoptotic function but still can inactivate the modified RIP1 and therefore, cause ripoptosme disassembly. In the presence of cFLIPS, procaspase-8 remains non-functional and so, active RIP1 accumulation which is followed by the recruitment of RIP3 brings about cell death via necroptosis [Reproduced from (Feoktistova et al., 2011) with a permission from the author].

54

1.3.4 Regulation of autophagy pathways by FLIPs Autophagy is a homeostatic process in which cytosolic materials such as proteins, organelles and pathogens are engulfed into double-membraned vesicles, termed the autophagosome, and then delivered to lysosome for degradation. Depending on the cellular conditions, the processed constituents can then be disposed of or recycled if needed. The latter provides cells with a mechanism of adaptation to starvation (Ryter et al., 2013). Autophagy can be also induced by factors such as hypoxia, and endoplasmic reticulum stress as well as exposure to a wide range of chemical and physical agents. Although autophagy is primarily a cytoprotective pathway, its uncontrolled upregulation can lead to cell death (Shintani and Klionsky, 2004). Too much or too little autophagy has been implicated in several human pathologies including neurodegeneration, myopathies, cancer, heart and liver disease (Mizushima et al., 2008). The core autophagic machinery comprises of the well-conserved family of autophagy-related (Atg) proteins. Molecular regulation of autophagy pathway occurs at three steps: (i) initiation, (ii) nucleation of an isolation membrane and (iii) elongation and completion of autophagosomes. Activation of ULK1 complex (ULK1, Atg13 and FIP200), which on resting conditions is inhibited by mTOR complex1, triggers the autophagy cascade. Next, the classIII PI3K complex (Beclin 1, hVps34, Atg14) mediates nucleation of autophagosomal membranes. The elongation and closure of the autophagosomal vesicles is dependent on the Atg5/Atg12/Atg16 complex and the ubiquitin-like protein, LC3. Unlike the first complex which dissociates from autophagosome once it is fully formed, LC3-phosphatidyl ethanolamine conjugates (aka LC-II) remain attached to vesicles. Attachment of phosphatidyl ethanolamine molecules to LC3 is mediated by Atg4 cysteine protease, Atg3 E2-like enzyme and Atg7 E1-like enzyme (Pyo et al., 2012). A screen conducted to identify KSHV proteins that could inhibit autophagymediated cell death, pulled out vFLIP. KSHV vFLIP, HVS vFLIP, MC159L, cFLIPL and cFLIPS all inhibit autophagy. The FLIP proteins interact with the Atg3, preventing LC3 conjugation to phosphatidyl ethanolamine and attachment to membranes for autophagic vesicle expansion (Lee et al., 2009) (Figure 1.7). During KSHV latency, deregulation of cell proliferation by vCyclin elicits DNA damage responses which, if unchecked, can promote autophagy and oncogene-induced senescence (OIS) (Koopal et al., 2007; Verschuren et al., 2002). By countering the v55

cyclin-induced autophagy, vFLIP plays an indispensable role in prevention of OIS, permitting the expansion of abnormal cells (Leidal et al., 2012). KSHV vFLIP mutants deficient in NEMO- or FADD-binding are capable of blocking autophagy (Lee et al., 2009; Leidal et al., 2012). Hence, the anti-autophagic role of vFLIP appears to be mediated by a unique motif, independent of its ability to induce NF-κB or block apoptosis. Indeed, Atg3-binding region of KSHV vFLIP was mapped to DED1 α2 helix (aa 20-29) and DED2 α4 helix (aa 128-139). Introducing peptides derived from these motifs can effectively block Atg3-vFLIP interaction without interrupting that of Atg3-LC3, leading to growth inhibition and autophagy-mediated cell death (Lee et al., 2009).

56

Figure 1.7. Cellular and viral FLIPs block autophagy. Autophagy is a homeostatic mechanism which is activated in response to various conditions such as absence of the amino acids, growth factors or oxygen. During starvation, mTORC1 that represses autophagy in normal conditions is inhibited which in turn causes ULK1 complex (ULK1, Atg13, FIP200) activation. Autophagy initiation is also dependent on other complex (Beclin 1, hVps34, Atg14) crucial for nucleation of autophagosomal membranes. The nucleation complex is activated through JNK1-mediated phosphorylation and subsequent inactivation of its inhibitory protein Bcl-2. The elongation and closure of the autophagosomal vesicles is dependent on the Atg5/Atg12/Atg16 complex and LC3. Unlike the first complex which dissociates from autophagosome once it is fully formed, LC3-phosphatidyl ethanolamine (PE) conjugates remain attached to vesicles. Attachment of PE molecules to LC3 is mediated by Atg3 E2-like enzyme and Atg7 E1-like enzyme. Cellular and viral FLIPs have been shown to supress autophagy by preventing the Atg3LC3 interaction.

57

1.3.5 Regulation of NF-κB pathways by FLIPs The ability of cFLIP proteins to activate NF-κB was initially demonstrated using overexpression experiments (Chaudhary et al., 2000; Hu et al., 2000). Later on, cFLIPL was found to mediate NF-κB activation following CD95 stimulation through a mechanism involving the recruitment of RIP1, TRAF1 and TRAF2 (Kataoka et al., 2000). This pathway is believed to depend on processing of cFLIPL to the p43 fragment as noncleavable mutant of cFLIPL does not induce it (Kataoka and Tschopp, 2004; Matsuda et al., 2014; Neumann et al., 2010). p43-FLIP has been found associated with NEMO after CD95 signalling (Neumann et al., 2010). One study reported that p22-FLIP, another cleavage product of cFLIPL/S/R can also directly interact with NEMO and activate IKK. p22-FLIP seems to be primarily important in NF-κB induction required for lymphocyte activation and DC maturation (Golks et al., 2006). Akin to its role in apoptosis, the expression level of cFLIPL is found to be a major determinant of its function as inhibitor or activator of DR-induced NF-κB signalling. Neumann et al showed that cFLIPL mediates CD95-induced NF-κB activation when expressed at moderate levels. On the other hand, high concentrations of cFLIPL can occupy the DISC, preventing the procaspase-8 activation, p43-FLIP production and NFκB activation (Neumann et al., 2010). This is in accordance with a number of other studies that propose cFLIPL downregulates CD95-induced NF-κB activity (Imamura et al., 2004; Kavuri et al., 2011; Kreuz et al., 2004). The reported inhibitory function of cFLIPS in TCR-induced NF-κB activation might be also explained by its capacity to effectively block procaspase-8 activation and the resultant generation of cFLIP cleavage products. Regulatory roles of cFLIP isoforms have been also controversially discussed in regard with TRAIL-induced NF-κB activation (Kavuri et al., 2011; Song et al., 2007; Wachter et al., 2004). While both positive and negative roles have been ascribed to cFLIPs in this context, a recent study argued against involvement of these proteins in TRAILinduced NF-κB signalling. Instead, FADD and caspase-8 were shown to mediate TRAILinduced signalling to IKK (Grunert et al., 2012). Among FLIP proteins, KSHV vFLIP is the most well-known inducer of NF-κB (Chaudhary et al., 1999; Chugh et al., 2005; Liu et al., 2002; Sun et al., 2003, 2006). Our group demonstrated direct, stable interaction between KSHV vFLIP and NEMO which leads to constitutive IKK complex activation (Field et al., 2003). Initially, it was suggested 58

that KSHV vFLIP binds to TRAF2 via a putative TRAF-binding motif and that this facilitates IKK activation by vFLIP (Guasparri et al., 2006). In contrast, later reports demonstrated that vFLIP could directly bind and activate NEMO independent of TRAF2 or TRAF3 (Matta et al., 2007). In KSHV-infected B cell lines, the IKK subunits alpha, beta and gamma, together with HSP90, are the only detectable proteins complexed with vFLIP; also NEMO immunoprecipitation depletes all detectable vFLIP from cell lysate (Field et al., 2003). KSHV vFLIP can also activate the alternative NF-κB pathway by upregulating the expression of p100 and its processing to p52. Non-canonical NF-κB induction by vFLIP is unique in that it does not require NIK and occurs through direct interaction between p100, vFLIP and IKK (Matta and Chaudhary, 2004). Both viral FLIPs from MCV (MC159 and MC160) block TNF-induced NF-κB signalling; MC159 has been shown to do this by binding to NEMO (Nichols and Shisler, 2006; Randall et al., 2012). On the other hand, MC160 is not detectable in association with any of IKK subunits. However, expression of MC160 leads to substantial decrease in levels of IKKα/β and interactions between these subunits, suggesting that MC160 may exert its inhibitory function by dissociating the IKK complex (Nichols and Shisler, 2006). The vFLIPs from RRV and HSV do not activate NF-κB, in contrast to vFLIP form EHV (Chaudhary et al., 1999; Lee et al., 2009; Ritthipichai et al., 2012). The interaction of these viral FLIPs with NEMO has yet to be studied. Detailed mechanisms of NF-κB activation by KSHV and cellular FLIPs will be discussed in chapters 3-5. 1.3.6 Regulation of MAPK pathways by FLIPs The MAPK signalling network is involved in fundamental cellular processes such as growth, proliferation, differentiation and apoptosis. This pathway is built upon a three-tier kinase module where a MAPK becomes activated through phosphorylation by a MAPKK which, in turn, is induced by an upstream MAPKKK (Dhillon et al., 2007). Six distinct groups of MAPKs have been described so far in mammalian cells: extracellular signalregulated kinase (ERK)1/2, ERK3/4, ERK5, ERK7/8, c-Jun N-terminal kinase (JNK)1/2/3 as well as the p38 isoforms α/β/γ [reviewed in (Arthur and Ley, 2013)]. While ERK1/2 are predominantly activated by growth factors, JNK and p38 kinases are induced in response to stress stimuli and cytokines (Pearson et al., 2001).

59

cFLIPL, cFLIPS and p43-FLIP can all activate ERK1/2 via recruitment of the upstream MAPK3, Raf-1 which activates MAPK2 MEKK and then ERK (Kataoka et al., 2000; Koenig et al., 2014). In activated T cells, high levels of cFLIP leads to simultaneous activation of NF-κB and ERK, promoting IL-2 secretion and T-cell proliferation (Kataoka et al., 2000). Nevertheless, a recent study reported that high expressions of cFLIP proteins can inhibit CD95-induced ERK activation, through blocking the procaspase-8 processing (Kober et al., 2011). It is possible that, like their role in apoptosis, cFLIPs differentially regulate ERK in cell-, pathway- and signal strength-dependent manner. Unlike ERK signalling, p38 and JNK appear to be preferentially inhibited by cFLIP isoforms. Gene silencing of cFLIP isoforms in keratinocytes enhances CD95- and TRAIL-induced p38 and JNK activation without affecting NF-κB activation (Kavuri et al., 2011). In addition, cFLIP−/− B-cells highly activate p38 and JNK signalling in response to LPS, while neither NF-κB nor ERK1/2 signalling is altered in these cells (Zhang et al., 2009b). Following TNFα activation, cFLIPL, but not cFLIPS, directly binds to MKK7, an activator of JNK (Nakajima et al., 2006). This stimulus-dependant binding inhibits the activation and interactions of MKK7 with its upstream kinases such as MEKK1, ASK1 (apoptosis-signal-regulating kinase1), and TAK1. As mentioned previously, cFLIPL is also targeted for degradation by JNK-induced E3 ubiquitin ligase Itch (Chang et al., 2006). Therefore, cFLIPL appears to operate as a molecular rheostat in TNFα-induced JNK signalling. While prolonged induction of JNK, which leads to cell death, is blocked by cFLIPL, its degradation by Itch prevents full blockade of the pathway. KSHV vFLIP was initially reported to activate JNK/AP-1 pathway in a TRAF2 dependent fashion, eventually leading to high expressions of cIL-6 (An et al., 2003). However, later studies failed to detect vFLIP-mediated JNK/AP-1 activation in either PEL or non-PEL cells (Sun et al., 2006; Ye et al., 2008). In marked contrast, vFLIP was demonstrated to suppress AP-1 activity -which is required for lytic KSHV replicationthrough NF-κB activation. This function of vFLIP is speculated to be central

for

maintaining the latent KSHV infection (Ye et al., 2008). 1.3.7 FLIPs as promising targets for anti-cancer therapies Enhanced cFLIP expression has been observed in a wide variety of cancers including melanoma (Yang et al., 2007), glioblastoma (Panner et al., 2009), colorectal 60

(Longley et al., 2006; Wilson et al., 2007), pancreatic (Haag et al., 2011; Kauh et al., 2010), ovarian (El-Gazzar et al., 2010; Park et al., 2009), prostate (Zhang et al., 2007) and gastric carcinomas (McLornan et al., 2010). Importantly, the higher levels of cFLIP correlate with more aggressive tumours (Korkolopoulou et al., 2004; Ullenhag et al., 2007; Valente et al., 2006; Valnet-Rabier et al., 2005; Wang et al., 2007a). In fact, cFLIP has been considered as a prognosis marker which may contribute to characterisation of patients at higher risks (Bagnoli et al., 2010). In support of this notion, increased amounts of cFLIPL were identified as an independent maker of adverse clinical outcome in several malignancies such as ovarian, colon and endometrial carcinomas as well as Burkitt’s lymphoma (Bagnoli et al., 2009; Safa et al., 2008). Abnormal upregulation of cFLIP proteins renders cancer cells resistant to not only CD95 (Mezzanzanica et al., 2004) and TRAIL-induced cell death (Geserick et al., 2008; Li et al., 2007) but also to chemotherapeutic drugs (Rogers et al., 2007). Resistance to TRAIL-induced apoptosis is particularly important as this type of cell death is found to be highly selective for the killing of neoplastic cells rather than normal ones (Walczak et al., 1999). This is shown to be due to an impaired expression of decoy receptors in cancer cells (Pan et al., 1997; Sheridan et al., 1997). Alike cFLIPs, vFLIPs play crucial roles in apoptosis evasion and therefore, oncogenicity of the associated viruses. Thus, targeting FLIP proteins appears as an attractive anti-cancer therapy, especially if combined with conventional approaches such as TRAIL treatment and chemotherapy. Sensitising malignant cells to apoptosis by inhibiting FLIPs may also allow for administering lower doses of chemotherapeutic agents, decreasing the drug-induced systemic toxicity in cancer patients (Safa and Pollok, 2011). Several strategic interventions have been utilised to inhibit FLIP variants which include (i) usage of compounds inhibiting transcription or translation of these proteins (ii) oligonucleotide- or RNAi-mediated silencing of FLIPs and (iii) targeting FLIPs for degradation. These therapeutic strategies and advantages and disadvantages of each method have been extensively reviewed in (Safa and Pollok, 2011; Shirley and Micheau, 2010).

61

1.4 HTLV-1 Tax: a functional analogue of the KSHV vFLIP 1.4.1 HTLV-1 biology Human T-cell lymphotropic virus type 1 (HTLV-1) was discovered in Japan where it shows a high prevalence, about 10% of the population are infected (Yoshida, 2005). The virus was isolated and sequenced by the Gallo lab at the National Cancer Institute in the US and was the first human retrovirus to be identified (Poiesz et al., 1980). In Japan it is mainly transmitted neonatally by breastfeeding (Matsuoka and Jeang, 2007), and although it mainly infects T-cells, it enters cells by binding to a widely expressed glucose transporter GLUT-1 (Manel et al., 2003). HTLV-1 causes an aggressive T-cell leukaemia, adult T-cell leukaemia (ATL), and is also associated with a progressive motor neuron disease known as HTLV-1 associated myelopathy/tropical spastic paraparesis (HAM/TSP) (Osame et al., 1987; Yoshida, 2005). ATL occurs with a long latency, up to 60 years, after infection with HTLV-1 (Matsuoka, 2003). Early in the disease infected T-cells are driven to proliferate by upregulation of the IL-2 receptor and secretion of IL-2 (Ballard et al., 1988; Wano et al., 1988). Then oligoclonal expansion of infected cells occurs. Finally, the aggressive leukemia develops, at this stage the cells have ceased to be dependent on autocrine IL-2; the transformed phenotype is dependent on secondary mutations, at least some caused by insertional mutagenesis induced by the viral genome (Grassmann et al., 2005). HTLV-1 is a complex retrovirus than encodes additional proteins along with those required for viral replication (Matsuoka and Jeang, 2007). The Rex protein is required for nuclear export of unspliced RNA, analogous to the Rev protein of HIV (Rimsky et al., 1988). 1.4.2 HTLV-1 Tax The Tax protein is responsible for the stimulation of T cell proliferation by HTLV1. Tax activates the NF-κB pathway (Ballard et al., 1988) and expression of Tax in T cells is sufficient to up-regulate both the IL-2 and its receptor, and drive cells to proliferate (Grassmann et al., 1989; Wano et al., 1988). Tax also upregulates HTLV-1 viral gene expression via recruitment of CREB to the LTR, enhancing viral replication (Yin and Gaynor, 1996; Zhao and Giam, 1992). Indeed Tax is a polyfunctional protein, interactions with a variety of host cellular proteins have been reported [reviewed in (Boxus et al., 2008)]. Later in infection Tax expression is lost and the HTLV-1 HBZ protein, encoded

62

by an antisense transcript, drives transformation; HBZ is also a transcriptional regulator with many opposite effects to Tax (Satou et al., 2006). Tax is of particular relevance to this study because, like some vFLIP proteins, it binds directly to NEMO to activate IKK (Chu et al., 1999; Harhaj and Sun, 1999; Jin et al., 1999), explaining its ability to constitutively activate NF-κB pathway leading to upregulation of NF-κB-dependent genes. Nevertheless, how the Tax–IKK physical association leads to IKK activation is incompletely understood. Two regions of NEMO, distinct from that recognised by KSHV vFLIP, have been reported to be necessary for Tax activation of IKK (See 3.1.2) (Xiao and Sun, 2000; Xiao et al., 2000). Tax also activates the alternative NF-κB pathway in a similar manner to vFLIP, by enhancing p100 processing, dependent on IKKα and NEMO, but not NIK (Xiao et al., 2001b). Therefore, these proteins from unrelated viruses, which share no amino acid homology, represent a remarkable example of convergent evolution in the manner in which they activate the NF-κB pathway.

1.5 Aims of the thesis Leading on from previous projects in our laboratory that demonstrated the necessity and dynamics of the vFLIP-NEMO interaction, I set out to achieve the following goals: 1) To identify the regions of NEMO required for activation of IKK by KSHV vFLIP, cellular FLIPs and HTLV-1 Tax 2) To determine how IKK is activated by KSHV vFLIP, cFLIPs and Tax

63

CHAPTER

2 2. MATERIALS AND METHODS

64

2.1 Materials 2.1.1 Molecular buffers and bacterial media All buffers and bacterial media were prepared in double distilled H2O (ddH2O). Table 2.1. Buffers and bacterial media. Buffer/media Phosphate-buffered saline (PBS)

pH 7.4

Tris-EDTA (TE)

8.0

Tris-acetate EDTA (TAE)

7.8

Elution buffer (EB) Luria Bertani broth

8.5 7.5

Luria Bertani agar Transformation buffer (TFB)-I

7.5 5.5

TFB-II

6.5

6x gel loading buffer

6.8

Composition 137 mM NaCl 2 mM KCl 10 mM sodium hydrogen phosphate (dibasic) 2 mM potassium hydrogen phosphate (dibasic) 10 mM Tris-HCl 1 mM EDTA 40 mM Tris-HCl 1 mM EDTA 20 mM sodium acetate 10 mM Tris-HCl 1% bacto-tryptone 0.5% bacto-yeast extract 10% NaCl LB broth plus bacto-agar 15g/L 30 mM potassium acetate 100 mM rubidium chloride 10 mM calcium chloride 50 mM magnesium chloride 15% glycerol acetic acid to desired pH 10 mM MOPS 75 mM calcium chloride 10 mM rhubidium chloride 15% glycerol KOH to desired pH 0.25% bromophenol blue 0.25% xylene cyanol FF 30% glycerol in water

65

2.1.2 Antibodies Table 2.2. Primary antibodies used for immunoblotting and immunoprecipitation. Target Atg3 Caspase-8 cFLIP† CYLD FADD FLAG epitope GAPDH GST HA epitope HOIL-1 HOIP IKKαβ MEKK3 NEMO pIKKα(S176/180)/ β(S177/181) pIKKα(S176)/β(S177) pIκBα (S32) pIκBα (S32/36) RIP1 SHARPIN

Clone

Manufacturer

Species/Isotype

Dilution

CST Santa Cruz Enzo Life Sciences

Rabbit, Polyclonal Goat, Polyclonal

1:1000 1:1000

Mouse, IgG1

1:2000

1/FADD M2

Sigma BD Sigma

1:1000 1:1000 1:2000

14C10

CST

Rabbit, Polyclonal Mouse, IgG1 Mouse, IgG1 Rabbit,

B-14

Santa Cruz

Y-11

Santa Cruz

H470 40/MEKK3

Walczack lab Sigma Santa Cruz BD

FL-419

Santa Cruz

16A6

CST

C84E11

CST

14D4

CST

12C2 38/RIP1

CST BD Walczak lab

C-20 NF6

Monoclonal Mouse, IgG1 Rabbit, Monoclonal Mouse, IgG2a Rabbit, Polyclonal Rabbit, Polyclonal Mouse, IgG1 Rabbit, Polyclonal Rabbit, Monoclonal Rabbit, Monoclonal Rabbit, Monoclonal Mouse, IgG1 Mouse, IgG2a Mouse, IgG1

1:4000 1:1000 1:1000 1:1000 1:2000 1:1000 1:1000 1:3000 1:1000 1:1000 1:1000 1:700 1:1000 1:1000

TAK1 M-579 Santa Cruz Rabbit, Polyclonal 1:1000 vFLIP 6/14 Collins lab Rat, Monoclonal 1:300 Manufacturers: BD, Beckton Dickenson Transduction Laboratories, CST, Cell Signalling Technology. † The anti-cFLIP(NF6) antibody has been raised against amino acids 1-194 of human cFLIP and therefore, recognises all of its isoforms and proteolytic fragments.

66

Table 2.3. HRP-conjugated secondary antibodies used for immunoblotting. Target

Host

Manufacturer

Dilution

Goat IgG

Rabbit

Dako

1:2000

Mouse IgG

Sheep

GE Healthcare

1:10000

Mouse IgG (ligh chain-specific)

Mouse

Jackson Immunoresearch

1:20000

Rabbit IgG

Pig

Dako

1:3000

Rabbit IgG (Fc-specific)

Goat

Jackson Immunoresearch

1:20000

Rabbit IgG (ligh chain-specific)

Mouse

Jackson Immunoresearch

1:20000

Rat IgG

Rabbit

Dako

1:1000

Fc: Fragment crystallisable region of an antibody. 2.1.3 Primers Table 2.4. Primers used for amplifying cDNAs of genes of interest. Amplicon

Target

Primer name

Primer sequence (5'→3')

BglII-cFLIP-FW

agatctgccaccatgtctgctgaagtcatccat

Vector cFLIPL cFLIPS p22-FLIP CYLD

pDual &

pCDNA3 cFLIPL-NotI-RS

gcggccgcttatgtgtaggagaggataagtttctttc

pDual &

agatctgccaccatgtctgctgaagtcatccat

BglII-cFLIP-FW

pCDNA3 cFLIPS-NotI-RS

gcggccgctcacatggaacaatttccaagaattt

pDual &

agatctgccaccatgtctgctgaagtcatccat

BglII-cFLIP-FW

pCDNA3 p22-NotI-RS

ggggcggccgctcaatccttgagac

pDual

ggatccgccaccatgagttcaggcttatggagccaagaa

BamHI-CYLD-FW

aaag FADD RIP1 HA-RIP1

pCAN pDual

CYLD-NotI-RS

gcggccgcttatttgtacaaactcattgttggactctgg

BamHI-FADD-FW

ggatccgacccgttcctggtgctgctgc

FADD-NotI-RS

gcggccgctcaggacgcttcggaggtagatg

AsiSI-RIP1-FW

gcgatcgcgccaccatgcaaccagacatgtccttgaatg

RIP1-NotI-RS

gcggccgcttagttctggctgacgtaaatcaagctgctc

pCDNA3 KpnI-HA RIP1-FW

ggtaccgccaccatgtacccatacgacgtcccagactac gctggtcaaccagacatgtccttgaatgtca

IκBα(1-54)

pGEX-

RIP1-NotI-RS

gcggccgcttagttctggctgacgtaaatcaagctgctc

BamHI-IκBα-FW

ggatccgccatgttccaggcg

IκBα∆54-EcoRI-RS

gaattctcagaggcggatctcctg

2T

67

Luciferase

pDual

BamHI-Luc-FW

ggatccaccgccatggaagacgccaaaaacataaagaa agg

NEMO

pDual

Luc-NotI-RS

gcgggcgcttacaatttggactttccgcccttcttggcc

AscI-SFFV-FW

ggcgcgccagtcctccgacagactg

SalI-NEMO∆254-RS

gtcgactcaccgctcactgcccaccacgctgctcttg

AscI-SFFV-FW

ggcgcgccagtcctccgacagactg

SalI-NEMO∆271-RS

gtcgactctagtcactcggcctgctggagctgctg

BamHI-vFLIP-FW

ggatccgccaccatggccacttacgag

∆254 NEMO

pDual

∆271

vFLIP

pDual &

pCDNA3 vFLIP-NotI-RS

gcggccgcctatggtgtatggcgatagtg

Table 2.5. Primers used for site-directed mutagenesis Mutation

Primer Sequence (5'→3')

cFLIP A56L

FW: tcggggacttgttggaactgctctac RS: gtagagcagttccaacaagtccccga

cFLIP K192/K195RR FW: gaatgttctccaagcagcaatccaaagaagtctcagagatccttcaaataacttcagg RS: cctgaagttatttgaaggatctctgagacttctttggattgctgcttggagaacattc cFLIP F114/L115AA

FW: tgtgcgagggatattaggtctttgatagctgcaagcaaggacactatagggtctc RS: gagaccctatagtgtccttgcttgcagctatcaaagacctaatatccctcgcaca

IKKβ S177/S181AA

FW: gctggatcagggcgctctttgcacagcattcgtggggac RS: gctggatcagggcgctctttgcacagcattcgtggggac

IKKβ S177/S181EE

FW: aggagctggatcagggcgaactttgcacagaattcgtggggaccctgc RS: gcagggtccccacgaattctgtgcaaagttcgccctgatccagctcct

RIP1 E620/D622AA

FW: gaaattgaccatgactatgcgcgagctggactgaaagaaaaggtt RS: aaccttttctttcagtccagctcgcgcatagtcatggtcaatttc

RIP1 G595/K596AA

FW: cccaatcagggaaaatctggcagcgcactggaaaaactgtgccc RS: gggcacagtttttccagtgcgctgccagattttccctgattggg

RIP1 K377R

FW: cccagcctgcagagtagactccaagacgaag RS: cttcgtcttggagtctactctgcaggctggg

vFLIP D102/E104RR FW: cactgttctccacgtacgcgggcggctgtgtgcgaggg RS: ccctcgcacacagccgcccgcgtacgtggagaacagtg vFLIP E104R

FW: ccacgtagacgggcggctgtgtgcgagg 68

RS: cctcgcacacagccgcccgtctacgtgg vFLIP F115/L116AA

FW: tgtgcgagggatattaggtctttgatagctgcaagcaaggacactatagggtctc RS: gagaccctatagtgtccttgcttgcagctatcaaagacctaatatccctcgcaca

vFLIP R12E

BamHI-vFLIP R12E- FW: ggatccgccaccatggccacttacgaggttct ctgtgaggtggcggagaaactgggcacgga

Table 2.6. Sequencing primers Primer

Sequence

SFFV FW

cgagctctataaaagagctca

pDual seq RS

taaagcagcgtatccacatagcgtaaaagga

pSIREN shRNAFW

atttcttgggtagtttgcag

pSIREN shRNA RS

gggctgctaaagcgcatgc

2.1.4 Plasmids used in this study Table 2.7. Mammalian and bacterial expression vectors used in this project. Plasmid

Origin

pDual (SFFV) NEMO WT (Ub) mCherry

Akira Shimuzu, Collins lab

pDual (SFFV) NEMO F312A (Ub) mCherry

Akira Shimuzu, Collins lab

pDual (SFFV) NEMO F238/D242RR (Ub) mCherry

Akira Shimuzu, Collins lab

pDual (SFFV) NEMO D242V (Ub) mCherry

this study

pDual (SFFV) NEMO ∆271 (Ub) mCherry

this study

pDual (SFFV) NEMO ∆254 (Ub) mCherry

this study

pDual (SFFV) vFLIP (Ub) GFP

Akira Shimuzu, Collins lab

pDual (SFFV) FLAG-Tax (Ub) GFP

Akira Shimuzu, Collins lab

pDual (SFFV) cFLIPL (Ub) GFP

this study

pDual (SFFV) cFLIPS (Ub) GFP

this study

pDual (SFFV) p22-FLIP (Ub) GFP

this study

pDual (SFFV) CYLD (Ub) GFP

this study

pDual (SFFV) RIP1 WT (PGK) GFP

this study

pDual (SFFV) RIP1 WT (PGK) GFP

this study

pDual (SFFV) RIP1 K377R (PGK) GFP

this study

pDual (SFFV) RIP1 G595/K596AA (PGK) GFP

this study

69

pDual (SFFV) RIP1 E620/D622AA (PGK) GFP

this study

pDual (SFFV) empty (Ub) GFP

David Escors, UCL

pSIN (SFFV) GFP

David Escors, UCL

pSIN (CMV) NF-κB luciferase

this study

pSIN (CMV) NF-κB mCherry

this study

pHIV-SIREN GFP

David Escors, UCL

pHIV-SIREN PuroR

Greg Towers, UCL

pHIV-SIREN HygroR

this study

pGL. IgK

this study

pGL. H2DK

this study

pRL.TK

David Guiliano, UEL

pNF-κB-H2DK-luciferase

Inna Lavrik, DKFZ, Germany

pCDNA3 empty

Pablo Rodriguez, UCL

pCDNA3 vFLIP

this study

pCDNA3 His-vFLIP

this study

pCDNA3 FLAG-vFLIP

this study

pCDNA3 vFLIP A57L

this study

pCDNA3 vFLIP F115/L116AA

this study

pCDNA3 vFLIP R12E

this study

pCDNA3 vFLIP R12E/E104R

this study

pCDNA3 vFLIP D102R/E104R

this study

pCDNA3 p22-FLIP WT

this study

pCDNA3 p22-FLIP A56L

this study

pCDNA3 p22-FLIP F114/L115AA

this study

pCDNA3 p22-FLIP K192/195RR

this study

pCDNA3 cFLIPS WT

this study

pCDNA3 cFLIPS A56L

this study

pCDNA3 cFLIPS F114/L115AA

this study

pCDNA3 cFLIPS K192/195RR

this study

pCDNA3 cFLIPL WT

this study

pCDNA3.1 FLAG-cFLIPL WT

Pascal Meier, ICR

pCDNA3 cFLIPL A56L

this study

pCDNA3 cFLIPL F114/L115AA

this study

70

pCDNA3 cFLIPS K192/195RR

this study

pCDNA3 cFLIP D196E/D376N

Inna Lavrik, DKFZ, Germany

pCDNA3 FLAG-Tax

this study

pCDNA3 Myc-NEMO WT

this study

pCDNA3 MycNEMO D242V

this study

pCDNA3 Myc NEMO F312A

this study

pCDNA3 Myc-NEMO F238/D242RR

this study

pCDNA3 HA-IKKα WT

this study

pCDNA3 HA-IKKα S176/S180AA

this study

pCDNA3 HA-IKKα S176/S180EE

this study

pCDNA3 HA-IKKβ WT

this study

pCDNA3 HA-IKKβ S177/S181AA

this study

pCDNA3 HA-IKKβ S177/S181EE

this study

pCDNA3.1 His-V5-HOIP

Henning Walczak, UCL

pCDNA3.1 FADD

Henning Walczak, UCL

pCDNA3 HA-FADD

this study

pCDNA3 HA-RIP1

this study

pGEX-2T IκBα (1-54)

this study

pGEX IκBα (1-73) S32/S36AA

Neil Perkins, Newcastle University

2.2 Molecular biology 2.2.1 Polymerase chain reaction (PCR) To amplify DNA fragments for subcloning into an expression vector, PCR reactions were performed using the high fidelity Phusion® DNA polymerase (NEB), as described in Table 2.8. Phusion DNA Polymerase possesses 3´→5´ proof-reading activity and generates blunt end products which, following purification, can be ligated into pJET1.2/blunt cloning vector. For colony screen PCRs, we used the goTaq® Green Master Mix (Promega, Madison, WI). The colony PCR is a convenient method to screen for the absence or presence of a plasmid DNA insert directly from the transformed E.coli colonies. This

71

method can also be used to determine the insert orientation following blunt-end DNA ligations. Table 2.9 summarises the reagents used for the PCRs with goTaq polymerase. Table 2.8. Phusion polymerase reaction mixtures Stock

Volume

Final

Concentration

added

concentration

-

to 50 µl

x5

10 µl

x1

dNTPs

10 mM

1µl

200 µM

Forward Primer

10 µM

2.5 µl

0.5 µM

Reverse Primer

10 µM

2.5 µl

0.5 µM

DNA template

100 ng

1 µl

100 ng

2 units/µl

0.5 µl

1 unit/50 µl PCR

Component Nuclease-free water Phusion HF† or GC† buffer

Phusion DNA polymerase Total volume

50 µl

†: As recommended by the manufacturer, HF buffer was used as the default buffer for high fidelity amplification, while GC buffer was used in PCRs with GC-rich template or those which did not work with HF. Table 2.9. goTaq polymerase reaction mixtures Component Nuclease-free water

Stock Concentration

Volume added

Final concentration

-

to 25 µl

-

x2

12.5 µl

x1

Forward Primer

10 µM

2.5 µl

1 µM

Revers Primer

10 µM

2.5 µl

1 µM

GoTaq Green Master Mix

DNA template

A small amount of E.coli colony

Total volume

25 µl

72

Reactions were run in a Hybraid thermal cycler using parameters listed in Table 2.10 and Table 2.11. Table 2.10. Cycling parameters for Phusion polymerase reactions Cycle step Initial Activation

Cycles

Temperature

Time

1

98 oC

30 seconds

98 oC

5-10 seconds

Denaturation Annealing

25-30

~ 5oC below Tm of primers

Extension

10-30 seconds

72 oC

15-30 seconds/kb

Final Extension

1

72 oC

5-10 min

Hold

1

4 oC

∞

Table 2.11. Cycling parameters for goTaq polymerase reactions Cycle step Initial Activation

Cycles

Temperature

Time

1

95 oC

2 min

o

Denaturation

95 C

30 seconds

o

Annealing

25-30

Extension

~ 5 C below Tm of primers

30 seconds

72 oC

1 min/kb

Final Extension

1

72 oC

5-10 min

Hold

1

4 oC

∞

2.2.2 Site-directed mutagenesis Mutations were introduced using either QuickChange® XL II Mutagenesis Kit (Agilent Technologies) according to the manufacturer’s instructions, or by overlap extension PCR method (OE-PCR). Both methods are based on amplification of the unmutated template DNA using complementary primer pairs both harbouring the desired mutation in their middle part (one forward and one reverse primer, here referred to as sense and antisense). The QuickChange XL-II mutagenesis relies on a single PCR reaction using mutation-containing primers and a plasmid DNA as template. Extension of the oligonucleotide primers by Pfu HF DNA polymerase generates a mutated plasmid 73

containing staggered nicks. DNA isolated from most strains of E.coli is dam methylated and therefore, susceptible to digestion with the restriction enzyme DpnI. Hence, treatment of the PCR mixture with DpnI results in digestion of the parental unmutated DNA; however, the mutated plasmid produced by the PCR is unmethylated and so, is preserved. The mutated plasmid can then be transformed into competent cells and amplified within bacteria. In OE-PCRs, three PCR reactions are performed to generate a DNA fragment with the desired mutation. For this method, apart from the mutation-harbouring internal primers, two primers (FW: forward and RS: reverse) one for each end of the final product are required. First, two PCR reactions using FW+antisense and sense+RS oligos were performed to generate two overlapping DNA segments containing the mutation of interest at their 3'- and 5'-terminal part, respectively. Finally, a third PCR was carried out to adjoin the two segments into a full-length product using FW and RS primers and products of the initial PCRs as template. These PCR products were then cloned into pJET1.2/blunt cloning vector. Presence of the mutation (and also absence of undesired sequence alterations) was verified by DNA sequencing, before subcloning the mutated genes into expression vectors. The primers used for site-direct mutagenesis were designed using the online QuickChange® Primer Design program; they are listed in Table 2.5. 2.2.3 Restriction digestions All restriction enzymes and their optimal buffers were purchased from Promega (Madison, WI) or New England Biolabs (NEB; Ipswich, MA). Digestion reactions for DNA analysis were normally performed at 37 oC for at least two hours, in a total volume of 10 µl (1 µl DNA, 0.5 µl of each enzyme, 1 µl of 10x buffer and 7 µl water). For isolation of plasmid backbone or inserts for ligation, reactions were carried out in a final volume of 30 µl (3 µl DNA, 1 µl of each enzyme, 3 µl of 10x buffer and 23 µl water). The incubation time was reduced to 30 min when using High-Fidelity restriction enzymes. To generate blunt-end DNA fragments from those with sticky-ends, the DNA was incubated with Klenow DNA polymerase at 37 oC for 30 minutes in a 30 µl reaction consisting of: 20µl gel-purified sticky end DNA, 1µl dNTP, 1 µl Klenow (NEB), 5µl water and 3 µl of x10 reaction buffer (10mM Tris-HCl, pH7.9, 50mM NaCl, 10mM MgCl2 and 1mM DTT). The Klenow is a proteolytic fragment of E.coli DNA polymerase which 74

retains polymerisation and 3'→5' exonuclease activity, but has lost 5'→3' exonuclease activity. Hence, it fills in 5' overhangs and chews back 3' overhangs, generating a bluntend DNA fragment (Jacobsen et al., 1974). 2.2.4 DNA ligations For ligation of DNA fragments, Rapid DNA Ligation Kit® (Thermo Scientific) was used. Ligation reactions were carried at room temperature for 5-10 minutes, in a total volume of 20 µl using 1 µl T4 DNA ligase (5 U/µl) and 4 µl of 5x Rapid ligation buffer. Next, 2µl of the reaction mixture was used for transformation of the competent cells. To minimise self-ligation of DNA fragments in reactions where both insert and vector were blunt-ended, we treated the backbone vector with calf intestinal alkaline phosphatase (CIAP) prior to ligation. This enzyme dephosphorylates 5'-ends of the vector, preventing them from binding to hydroxyl groups of the 3'-ends and therefore, minimising the re-circulation of the cut vectors. The dephosphorylation reactions were performed using 1µl of CIAP (1 U/µl, Promega) at 37 oC for 15 min, followed by further 15 min incubation at 56 oC to heat-inactivate the enzyme. 2.2.5 Annealing DNA oligonucleotides for subcloning into a plasmid In order to introduce a short stretch of DNA into a plasmid, we designed overlapping oligonucleotides that once assembled could be directly cloned into the overhangs generated by restriction digest of the destination vector. For annealing the oligonucleotides, 2.5 µl of each (reverse and forward) form a 100 µM stock were mixed and diluted in 50 µl of annealing buffer (10mM Tris pH 8.0, 50mM NaCl and 1mM EDTA). The mixture was incubated at 95 oC for 5 min and then, allowed to cool down to room temperature on the bench for approximately 45 min. The annealed oligonucleotides were diluted 10 times using nuclease-free water. Finally, the annealed oligonucleotides were mixed with cut vector in ratios between 3:1 to 6:1, in a standard ligation reaction as described in section 2.2.4. 2.2.6 Agarose gel electrophoresis and recovery of DNA Agarose gel electrophoreses were performed to separate, characterise and purify the DNA samples. DNA fragments were electrophoresed in 1% agarose gels (Invitrogen, Carlsbad, CA) containing 5 µg/ml ethidium bromide (Dutscher Scientific, Essex, UK), using TAE buffer as running buffer. Prior to loading, DNA samples were mixed with x6 75

TriTrack DNA loading dye (Thermo Scientific) and then, were run alongside 1kb Plus DNA ladder (Invitrogen, Carlsbad, CA) for the estimation of band sizes. Gels were finally visualised under UV illumination and analysed using the GeneSnap image acquisition software (SYNGENE, NJ, USA). For DNA purification after electrophoresis, the bands of interest were excised form the gel and purified using Qiaquick Gel Extraction kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. 2.2.7 Preparation, transformation and growth of competent bacteria For the generation of competent cells, we used the rhubidium chloride method. Briefly, XL10 gold E.coli (Agilent were streaked onto an LB-agar plate containing Tetracycline (10 µg/ml) and incubated overnight at 37 oC. A single colony was then picked and cultured overnight in 10 ml of LB medium with Tetracycline (10 µg/ml). Next, 100 ml of LB medium without antibiotics was inoculated with 1 ml of overnight culture and grown at 37 oC for approximately 90-120 min, until an OD600 of 0.5-0.6 was achieved. The bacterial suspension was then cooled on ice for 10 min and harvested at 3000 RPM and 4 °C for 10 min. The pellet was gently suspended in 50 ml of TFB-I buffer (See table 2.1) and left on ice for 5 min. Following this incubation, cells were spun down again as explained above, re-suspend in 4 ml of ice-cold TBF-II buffer and left on ice for additional 15 min. Finally, aliquots of 100 µl were prepared and stored at -80 oC until use. For transformation, vials of competent bacteria were thawed on ice and inoculated with 100-500 ng of plasmid DNA or 2-5 µl of ligation reaction. After incubation on ice for 20 minutes, the bacteria were heat-shocked for 15 seconds at 42 oC (or 2 minutes at 37 ºC) and put back on ice for an additional 2 minutes. The transformed cells were then streaked onto LB-agar plates containing selection antibiotics (Ampicillin 50 µg/ml, or Kanamycin 30 µg/ml) and grown overnight at 37oC. 2.2.8 DNA purification and quantification Single colonies of transformed bacteria were picked from LB agar plates and grown overnight at 37 ºC, in 4 ml (minipreps) or 100 ml (midipreps) of LB broth supplemented with appropriate selection antibiotics (usually Ampicillin 50 µg/ml). Plasmid DNA was purified using QiaPrep Spin Miniprep or Plasmid Midiprep kits (Qiagen) following the 76

manufacturer’s protocol. The purified DNA was normally eluted in EB buffer or nuclease-free water and quantified using NanoDrop® 3300 spectrophotometer (Thermo Scientific, Wilmington, DE). 2.2.9 Extraction of cellular RNA and cDNA synthesis The cDNAs encoding human IKKα and IKKβ, were amplified from cDNA synthesised from Jurkat T cells. To do this, we first isolated total cellular RNA using RNeasy mini kit (Qiagen) according to the manufacturer’s instructions. Next, mRNA was reverse transcribed to cDNA using the qScript® II cDNA SuperMix (Quanta BioSciences). Reverse transcription-PCR reaction mixtures and parameters are shown below in Table 2.12. Finally, 2 µl of the cDNA/RNA sample was used as template to amplify IKKα and IKKβ cDNAs by means of a standard PCR reaction. Table 2.12. RT-PCR reaction mixtures (top) and thermocycler parameters (bottom). Component

Volume added

qScript cDNA SuperMix

4 µl

RNA template

1 µg (whatever the volume)

Nuclease-free water

to 20 µl

Total volume

20 µl

Cycle

Temperature

Time

Annealing

25oC

5 min

Elongation

42oC

30 min

Denaturation

85oC

5 min

Hold

4o C

∞

2.2.10 DNA sequencing DNA sequences were verified, using standard or customised primers, at the University College London or Beckman Coulter sequencing services. The list of customised primers is given in Table 2.6.

77

2.3 Tissue culture HEK293T cells were used for viral packaging, lentivector titrations and also testing of the transgene expression of cloned plasmids. They are highly transfectable cells derived from human embryonic kidney cells which stably expresses the large T antigen from simian virus 40. The expression of T antigen contributes to episomal maintenance and increased replication of vectors containing SV40 origin/promoter (DuBridge et al., 1987; Pear et al., 1993). This makes HEK293T ideal target cells for viral preparations, gene expression and protein production. All adherent growing cell lines including HEK293T, HEK293 and mouse embryonic fibroblasts (MEF) were cultured in Dulbecco’s modified Eagle’s medium (DMEM)(Sigma), supplemented with 10% foetal bovine serum (FBS)(Gibco, Paisley, UK), 2 mM L-glutamine (Gibco), 100 U/ml Penicillin and 100 µg/ml Streptomycin (Gibco). These cells were grown at 37 oC in a 10% CO2 incubator and passaged every 2-3 days, using Trypsin/EDTA (Sigma). The wild-type and MEKK3 KO MEFs were kindly provided by Prof. Philip Cohen (University of Dundee, Scotland). Jurkat SVT35 cells and the 70Z/3 Pre-B cell line were cultured in Roswell Park Memorial Institute (RPMI) medium, supplemented with 10% FBS, 2 mM L-glutamine, 100 U/ml Penicillin, 100 µg/ml Streptomycin and 50 µM 2-Mercaptoethanol (Gibco). These cells were grown in a 5% CO2 incubator and passaged 1:10 (70Z/3) or 1:5 (Jurkat) every 3 days. The 70Z/3 and Jurkat SVT35 cells as well as their respective NEMOdeficient derivatives, 1.3E2 (Yamaoka et al., 1998) and JM4.5.2 cell (Harhaj et al., 2000), were kind gifts of Prof. Alain Isräel (Pasteur institute, Paris, France). All cell lines were frozen in FBS containing 10% dimethyl sulfoxide (DMSO) at a density of 4-8 x106 cells/ml. Cell counting was performed either manually by use of a Neubauer chamber slide, or by using the Muse® cell analyser (Millipore).

2.4 Lentivectors 2.4.1 Lentiviral transfer plasmids In this study, the pDual promoter lentivector backbone which has been previously described by our group (Arce et al., 2009; Rowe et al., 2009), was used. This construct incorporates two expression cassettes controlled by the spleen focus-forming virus (SFFV) and ubiquitin (Ub) promoters (Figure 2.1A). The SFFV promoter drives the 78

expression of a target gene (e.g, vFLIP or NEMO) while the Ub promoter initiates the expression of a fluorescent protein (emerald green fluorescent protein (eGFP) or mCherry). The cDNA template of transgenes were amplified using reverse and forward primers with incorporated 5' BamHI and Kozak sequences (forward) and 3' NotI site (reverse), and then were subcloned into BamHI-NotI restriction sites under the control of the SFFV promoter. If the BamHI site was present within the gene of interest, the BglII site was used which gives compatible cohesive ends. If this was not applicable either, a pDual vector encoding AsiSI and NotI insertion sites (provided by Dr. Gary Britton), was used. Sequences of primers used for amplification of the transgenes have been listed in Table 2.4.

cDNA sources cDNA templates were from the following sources: p22-FLIP and non-cleavable cFLIPL (Prof. Inna Lavrik, DKFZ, Germany), WT cFLIPL (Prof. Pascal Meier, ICR), RIP1 and FADD (Prof. Henning Walczak, UCL), CYLD (Addgene plasmid depositary), mCherry and firefly luciferase (Dr. David Escors). E.coli expressing plasmids encoding human cFLIPS cDNA (Clone ID: IHS1380-212917810) was obtained from Open Bioystems (Thermo Scientific). Human IKKα and IKKβ cDNAs were isolated from a cDNA pool synthesised from Jurkat cells. 2.4.2 Vector production Lentivectors (LV) were generated by a 3-plasmid transient co-transfection of HEK293T cells using a transfer vector, a 2nd generation HIV-derived packaging plasmid (p8.91) and a plasmid expressing vesicular stomatitis virus-glycoprotein (VSV-G) envelope (pMD.G). The p8.91 and pMD.G vectors were obtained from Plasmid factory (Bielefeld, Germany) and have been described previously (Demaison et al., 2002) (Figure 2.1B and C). One day prior to transfection, 107 HEK293T cells were seeded in 15 cm2 dishes to reach a cell confluency of 70-80% the following day. Cells were then transfected using the FuGENE® HD transfection reagent (Promega) with the following mixture of components (per plate):

79

Table 2.13. Components of transfection mixture for LV production (amounts for a 15 cm plate) Reagent

Quantity

LV transfer plasmid

3.75 µg

p8.91

2.5 µg

pMD.G

2.5 µg

FugeneHD

45 µl

Opti-MEM (Gibco)

500 µl

The mix was incubated at room temperature for 20 minutes and then added dropwise to cells in fresh DMEM. Twenty-four hours later the medium was changed and subsequently, three collections were made at day 2, 3 and 4 post-transfection. Supernatants containing lentiviral particles were then passed through 0.45 µm filters and concentrated 200-fold by ultra-centrifugation (20,000 RPM, at 4 oC for 2 hours) in a Sorvall centrifuge (Beckman Coulter). Viral pellets were then resuspended in ice-cold whole medium, incubated on ice for 1-2 hours and finally, aliquoted and stored at -80 oC, until use. 2.4.3 Viral titrations Titration of LVs expressing a fluorescent protein (usually GFP or mCherry) was performed using flow cytometry technique, while for those lacking a fluorescent protein gene, quantitative PCR (qPCR) method was used.

2.4.3.1 LV titrations by flow cytometry 2x105 HEK293T cells were seeded in 24-well plates in 0.5 ml of DMEM, three to four hours before transduction. Cells were then transduced with serial dilutions (1 in 5, 25, 125, 625 and 3125) of the prepared lentiviral particles. Twenty-four hours later, plate wells were topped up with an additional ml of medium. Three days following transduction, cells were harvested, washed in PBS twice, pelleted in FACS tubes (1,500 RPM, at 4oC for 5 min) and resuspended in 300 µl of PBS. Subsequently, the percentage of fluorescent protein-expressing cells was quantified by flow cytometry using LSRFortessa® cell analyser (BD Biosciences) and BD Diva software. Dilutions which resulted in 10-30% transduction rate –which fall within the linear range of the titration graph- were used to calculate virus titrations using the following formula: 80

# ×    % ×     × 1000 =   /     

2.4.3.2 LV titrations by qPCR This titration method calculates the number of integrated vector copies per cell. 2x105 of HEK3293T cells were plated out in 24-well plates and transduced with 2-5 µl of LV. Three days post-infection, genomic DNA was extracted using DNeasy® Blood and Tissue kit (Qiagen) and quantified by Nanodrop. Standards containing 101 to105 copies of pHV per µl were prepared. We used primers and fluorescent Taqman probe which recognise the HIV leader sequence. Reactions were performed in duplicates using Realplex Eppendorf mastercycler. The PCR program included an initial step of 90 oC for 10 min, followed by 40 cycles of 90 oC for 1 min and 60 oC for 10 min. Finally, the number of cells was estimated from extracted gDNA concentration and then, based on the standard curve values, number of integrant copies per cell was calculated. The oligonucleotide sequences and the amount of reagents used for each reaction have been summarised in the following table. Table 2.14. Components of qPCR reaction for titration of lentivectors. Reagent

Quantity/rxn

Sequence

FW primer (GT248)

0.5 µl

5’ tgtgtgcccgtctgttgtgt 3’

RS primer (GT 249)

0.5 µl

5’ gagtcctgcgtcgagagagc 3’

Strong stop probe

0.25 µl

FAM 5’cagtggcgcccgaacaggga3’ TAMRA

Quantitect Master

12.5 µl

Mix ddH2O Sample (Std or

8.25 µl 2 µl

unkown) Total

25 µl

81

2.4.4 Cell transductions Unless otherwise stated in the figures, cells were infected with lentiviral vectors at the following multiplicity of infection (MOI): HEK293/293T (10), Jurkat cells (20), MEFs (50) and 70Z/3 cells (100). At least 24 hours were allowed for the transgene expression before performing the functional assays.

Figure 2.1. Lentiviral constructs. A) pDual vector. B) 2nd generation packaging plasmid. C) Envelope-expressing vector. D) pGIPZ. E) pHIV-SIREN. CMV: cytomegalovirus, cPPT: central polypurine tract, HygroR: hygromycin resistance gene, IRES: internal ribosome entry site, LTR: long terminal repeat, PuroR: puromycin resistance gene, RRE: rev response element, SFFV: spleen focus-forming virus, tGFP: turbo GFP, Ub: ubiquitin. VSV-G: vesicular stomatitis virus-glycoprotein, WPRE: woodchuck hepatitis virus post-transcriptional regulatory element, Ψ: packaging signal.

82

2.5 Generation of stable knock-down cell lines Short hairpin RNA (shRNA)-expressing lentivectors were produced using pGIPZ lentiviral plasmids (UCL Open Biosystems shRNA library). Alternatively, custom shRNAencoding sequences were designed and cloned into pHIV-SIREN lentivector backbone (modified from pSIREN RetroQ, Clonetech). 2.5.1 pGIPZ pGIPZ plasmids contain CMV promoter driving the transcription of turboGFP marker gene which is followed by an IRES (Internal Ribosome Entry Site), Puromycin resistance gene and an shRNA-encoding sequence within a microRNA30 context (Figure 2.1D). HEK293T cells were transduced with shRNA lentivectors (normally 5-7 targeting viruses for each protein) at an MOI of 50 and 48 hours later, Puromycin was added to medium at 1µg/ml concentration to select infected cells. A non-silencing and a GAPDHtargeting lentivectors were produced and tested alongside gene-specific shRNAs as negative and positive controls, respectively. Finally, the knockdown of protein expression was analysed by immunoblotting. 2.5.2 pHIV-SIREN If pGIPZ constructs did not result in a successful knockdown or a different selection marker was needed, customised shRNA expressing oligonucleotide pairs were designed using Clonetech’s online shRNA target finder and shRNA designer softwares and then, subcloned into the pHIV-SIREN backbone. Upon hybridisation, oligonucleotides designed by this software produce 5' EcoRI and 3' BamHI sticky ends which can be then ligated into the compatible sites within the destination vector. The pHIV-SIREN is a bicistronic lentiviral construct designed to express a shRNA using the human U6 promoter and a selection marker gene (GFP, or Puromycin, or Hygromycin) driven by the phosphoglycerate kinase (PGK) promoter (Figure 2.1E). Selection of cells with HygromycinB (HygroGold®, Source Biosciences) was performed for two weeks at a concentration of 200 µg/ml. Cells transduced with GFP-expressing LVs were isolated by fluorescence-activated cell sorter (FACS).

83

2.5.3 shRNA sequences Table 2.15. shRNA-targeted sequences. Targeted Gene

Ref Seq

Atg3 (Hs)

NM_022488.4 1

Caspase-8 (Hs)

FADD (Hs)

shRNA#

Clone number

Sequence (5'→3')

V2LHS_201190

cattgaccatattcatcaa

●

2

V2LHS_14033

ggacttatatgtttatgca

●

3

V2LHS_11836

cagccttacttgtttaata

4

V3LHS_319838

aagctgtcattccaacaat

5

V3LHS_319835

tggaaaataaggacaatat

NM_001228.4 1

V2LHS_112731

gtcatcctgggagaaggaa

2

V2LHS_112733

cctactttcacactaagaa

3

V2LHS_112730

gagctgctcttccgaatta

4

V3LHS_320399

gggtggttattgaaagtaa

●

5

V3LHS_398368

tgcacagtagagcaaatct

●

6

V3LHS_398369

gcaacaaggatgacaagaa

NM_003824.3 1

V2LHS_16975

caggacgaattgagataat

2

V2LHS_16973

cgggatctcgtatctttaa

3

V3LHS_388202

ccgatgtcatggaactcag

4

V3LHS_412699

tgcaattctacagtttctt

5

V3LHS_388204

aggctggctcgtcagctca

6

V3LHS_388203

tgaccgagctcaagttcct

7

V3LHS_412698

aagatcttgtctccactaa

●

●

HOIL-1 (Hs)

NM_006462.4

pHIV-SIREN

ccacaacactcatctgtcaaa

●

HOIP (Hs)

NM_017999.4 1

V3LHS-399213

agctgctgtgctatgttcc

●

2

V3LHS-399214

agcacggaggtgatgtgtc

3

V3LHS-399210

tggcgtggtgtcaagttta

4

V3LHS-399212

tgccgagtgatagagcaga

NM_203351.1 1

pHIV-SIREN

cctggatatgagaccatga

2

pHIV-SIREN

gagtgacgtcagaatcaag

3

V2LHS_1970

gcattgaactcaatcatga

4

V2LHS_151699

caaatgtcatgctgcctta

5

V2LHS_151695

ctgaccatcttcatggagt

6

V3LHS_329522

tcggtgaaagaccagttga

7

V3LHS_329524

agacacaggtcactcaaat

MEKK3 (Hs)

84

●

●

RIP1 (Hs)

8

V3LHS_329521

accaagatgatcttgataa

9

V3LHS_329520

agaatcaagttcgagcaca

NM_003804.3 1

V2LHS_241668

cagttgataatgtgcataa

2

V2LHS_17422

accaacagatgaatctata

3

V3LHS_638037

agagtaaactccaagacga

4

V3LHS_638038

acagagaagtcggatgtgt

5

V3LHS_638034

agctgctaagtaccaagct

6

V3LHS_638036

tgggcgatatttgcaaata

7

V3LHS_645251

tgaagtggacgcctcactt

8

V3LHS_638040

accactagtctgacggata

9

V3LHS_340514

gcgagatggactgaaagaa

1

pHIV-SIREN

gtgttctcagagctcggtttc

2

pHIV-SIREN

ctgtccttcctgcaccttcat

3

pHIV-SIREN

tggaaacttgacggagaga

NM_003188.3 1

V2LHS-153758

cagatgagccattacagta

2

V2LHS-201931

cctttggagttgtttgcaa

3

V2LHS-201612

cgcccttcaatggaggaaa

4

V2LHS-202210

cgtgtgaaccatcctaata

5

V3LHS-370925

agagtgaatctggacgttt

6

V3LHS-370924

agcttttagatgaaaacaa

7

V3LHS-370923

aggtagtaattacagtgaa

NM_172688.3 1

pHIV-SIREN

cctgaacttcgaagagatc

2

pHIV-SIREN

cgaagagatcgactacaag

3

pHIV-SIREN

ctgagaggaaggctttcat

4

pHIV-SIREN

gcctgaatccagtatgtct

●

ccctcatttcctggtatgacaa

●

atctcgcttgggcgagagtaa

●

SHARPIN (Hs)

TAK1 (Hs)

TAK1 (Mm)

GAPDH (Hs)

NM_0012897

●

●

●

●

●

●

45.1 non-silencing

Hs: Homo sapiens, Mm: Mus musculus, RefSeq#: Reference sequence number ● indicates the shRNAs resulting in the highest knockdown efficiency.

85

2.6 Western blotting Cells were washed twice with ice-cold PBS and then lysed by adding PBS-T lysis buffer (PBS containing 1% Triton X100 and 5% Glycerol, supplemented with complete protease inhibitor cocktail (Roche Diagnostics)) and incubating on ice for 10-15 min. After this, whole cell lysates were clarified by centrifugation at 13,000 RPM, 4 oC for 20 minutes. The supernatants were transferred to new tubes and quantified for total protein concentration using bicinchoninic acid (BCA) colorimetric assay (Pierce® BCA protein assay kit, Thermo Scientific, Rockford, IL). After adjusting samples for equal protein concentrations, they were mixed with 4x Laemmli sample buffer and boiled at 95 oC for 5 minutes. 30-50 µg of the prepared samples were subjected to electrophoresis in a 4% SDSpolyacrylamide stacking gel, followed by an 8-12% SDS-polyacrylamide separating gel at 120 Volts for 2 hours. The separated proteins were transferred to polyvinylidene difluoride (PVDF) membrane using semi-dry transfer system and subsequently blocked with 5% skimmed milk in PBS-T (PBS with 0.1% Tween20) for an hour. Next, it was incubated overnight at 4 oC with appropriate dilution of a primary antibody in PBS-T containing 5% BSA and 0.1% Na3N. The membrane was washed in PBS-T five times for 5 minutes and incubated with horseradish peroxidase (HRP)-conjugated secondary antibody in PBS-T with 5% non-fat milk for an hour at room temperature. After thoroughly washing the membrane as stated above, it was incubated with LumiGLO® ECL (Cell Signalling Technology) for a minute and then, drained and exposed to Hyperfilm ECL (Amersham). For reprobing with another antibody, membrane was incubated with stripping buffer for up to 45 minutes at 55 oC to denature both primary and secondary antibodies. This was followed by multiple washing steps with dH2O and PBS-T. Membrane was then blocked for one hour and the immunoblotting procedure was continued as explained above. The buffers and antibodies used for immunoblotting have been summarised in Table 2.14 and Table 2.2, respectively.

86

Table 2.16. Western blot buffers and gel reagents Buffer/gel

Composition

Laemmli sample buffer

50 mM Tris, pH6.8 10% Glycerol 2% SDS 5% 2-Mercapthoethanol 0.2 mg/ml Brumophenol blue 0.1 M DTT

Running buffer

25 mM Tris, pH8.5 200 mM Glycine 0.1% SDS

Transfer buffer

100 mM Tris, pH6.8 200 mM Glycine 20% Methanol

Stripping buffer

62.5 mM Tris-HCl, pH6.8 2% SDS 100 mM 2-Mercaptoethanol

4% stacking polyacrylamide gel

125 mM Tris-HCl, pH6.8 4% Acrylamide/bis 10% SDS 0.1% TEMED 1% Ammonium persulphate (APS)

10% separating polyacrylamide gel

125 mM Tris-HCl, pH8.8 10% Acrylamide/bis 10% SDS 0.1% TEMED 1% APS

2.7 Immunoprecipitation After washing twice with ice-cold PBS, cells were lysed with PBS-E lysis buffer (PBS containing 1% TritonX100, 2 µM EDTA and complete Protease inhibitor cocktail) and incubated on ice for 15 minutes. The cell lysate were clarified as mentioned above and pre-cleared by incubating with beads for an hour at 4 oC. The pre-cleared lysate was 87

incubated with 1:100 dilutions of the desired antibody (test or isotype-matched control) plus 20 µl of ProteinA or G slurry beads at 4 oC for 2 hours while rotating. Next, beads were pelleted by centrifugation at 2,000 RPM for 2 min and washed twice with the high salt wash buffer (recipe in Table 2.17) followed by two additional washes with the PBS-E solution. Immune complexes were dissociated by adding 20 µl of 2x Laemmli buffer and boiling at 95 oC for 5 minutes. Finally, co-immunoprecipitation results were analysed by western blotting.

2.8 In vitro IKK kinase assay Cytoplasmic extracts were prepared using kinase lysis buffer (see Table 2.17) and quantified for protein concentration using BCA method. 500-1000 µg of lysates were incubated with 20 µl Protein A sepharose beads and 1 µg anti-NEMO (sc-8330, Santa Cruz) for at least 2 hours at 4 oC. Beads were then washed in 0.5 ml of high salt wash buffer three times, 5 minutes each, followed by two more washes with kinase wash buffer. After removing all the residual wash buffer, 40 µl kinase reaction buffer plus 2 µg GSTIκBα (1-54) and 0.5 µl p32γ-ATP was added to each sample and tubes were immediately incubated at 30 oC for 30 min. Reactions were stopped by adding 15µl 4x Lammeli buffer and incubating the tubes at 95 oC for 5 min. The samples were separated by 10% SDSPAGE which was then dried out on a filter paper and exposed to a storage phosphor screen (GE Healthcare) for 15-30 minutes. To visualise the radiolabelled bands, the storage phosphor screen was scanned using Typhoon PhosphorImager (GE Healthcare) and the intensity of bands were quantified by ImageQuant TL 7.0 software (GE Healthcare). Table 2.17. Composition of kinase assay buffers. Buffer

Composition

Kinase lysis buffer

20 mM Tris-HCl, pH7.5 150 mM NaCl 1% Triton X-100 5% Glycerol 1 mM PMSF Protease inhibitor cocktail

High salt wash buffer

25 mM Tris-HCl, pH7.6 500 mM NaCl

88

1 mM EGTA 1 mM DTT 1% Triton X-100 5% Glycerol 1 mM Na3VO4 10 mM β-glycerophosphate 5 mM NaF 1 mM PMSF Kinase wash buffer

20 mM HEPES, pH7.6 50 mM NaCl 20 mM β-glycerophosphate 0.5 mM DTT 1 mM PMSF

Kinase reaction buffer

20 mM HEPES, pH7.6 50 mM NaCl 10 mM MgCl2 2 mM DTT 20 µM ATP 0.1 mM Na3VO4

2.8.1 Expression and purification of GST-IκBα(1-54) First,

IκBα(1-54)

was

PCR-amplified

using

the

primers

FW:

5'ggatccgccatgttccaggcg3' and RS: 5'gaattctcagaggcggatctcctg3', containing BamHI and EcoRI restriction sites (underlined), respectively. The PCR product was cloned into pJET1.2, sequence-verified and then digested and subcloned into the pGEX-2T vector (kindly provided by Dr. Pablo Rodriguez, UCL Cancer Institute). Next, BL21 bacteria (NEB) were transformed with pGEX-2T-IκBα, streaked onto ampicillin LB-agar plates and grown overnight. A single colony was picked and grown at 37 °C in 100mL LB-broth containing 50 µg/ml ampicillin and 2% glucose until an absorbance of 0.4 to 0.5 at 600nm was reached. The bacterial culture was then induced by addition of 2mM isopropyl-β-thiogalactopyranoside (IPTG) for 3 hours at 37 °C. To act as a negative control, 5 ml of culture was be removed before induction and grown along with the induced culture. After incubation time, cultures were harvested by centrifugation 89

at 4,000 RPM for 10 minutes. The pelleted bacteria were then resuspended in B-PER Bacterial Protein Extraction Reagent (Pierce, ThermoSceintific), at 4 ml per gram of bacterial pellet, to lyse and release the GST fusion protein. Subsequently, the lysates were ultra-centrifuged at 20,000 RMP for 20 min at 4 oC to remove the bacterial debris. The supernatant was removed for purification with the pellet being resuspended in PBS for SDS-PAGE analysis after protein purification. The supernatant was transferred to a 1 ml Glutathione Spin Column (Pierce) and the GST-IκBα was purified following the manufacturer’s instructions (the protocol states that 3 fractions of eluted protein are produced with a possible 10 mg of protein being purified). Following the elution of the GST-IκBα from the spin column, the elution fractions, lysate, lysate supernatant and flow-through samples were subjected to SDSPAGE and coomassie stained to determine the presence of the protein (Figure 2.2) After confirming the presence of GST-fusion proteins, the samples were dialysed to remove glutathione and other sample contaminants, using the 10 kDa molecular weight cut-off Vivaspin® columns (Millipore) and a dialysis buffer (20mM Tris-HCl, pH8.0, 100mM NaCl, 0.2mM EDTA, 10mM β-glycero-3-phosphate and 10% glycerol). Finally, the protein samples were adjusted to a concentration of 1 µg/µl and stored at -80 oC until use.

Figure 2.2. Expression and purification of the GST-IκBα.

90

2.9 In vitro IκBα phosphorylation assay Resting HEK293T cells were lysed in HTX lysis buffer (10mM HEPES pH7.4, 10mM MgCl2, 1mM MnCl2, 0.1mM EGTA, 0.5% Triton X100) supplemented with complete protease inhibitor cocktail (Roche) and the S100 fractions were prepared by sequential centrifugation at 1,000 and 100,000g for 5 and 45 minutes, respectively. The cleared lysates were then incubated with recombinant FLIP and ATP generating system (10x stock: 10mM ATP, 20mM HEPES pH7.2, 10mM MgCl2, 350mM creatinine phosphate and 500 µg/ml creatinine kinase ) for 10 min at 37 oC. The reactions were terminated by adding 5x Laemmli buffer and boiling at 95 oC. Eventually, rec-FLIPinduced IKK activation rate was detected by immunoblotting for downstream indicators such as pIkBα and pIKKα/β. Levels of total IKKα/β served as control for equal protein loading. 2.9.1 Purification and expression of recombinant vFLIP, p22-FLIP and GB1p22FLIP Recombinant vFLIP was overexpressed and purified as described in Bagnéris et al (Bagnéris et al., 2008). For the production of recombinant p22-FLIP (and GB1-p22FLIP), cFLIPS was first PCR amplified from a human thymus cDNA library (Clontech) using the primers 5'ttgctagcatgtctgctgaagtc3' and 5'ttctcgagtcacatggaacaatttc3' containing NheI and XhoI restriction sites respectively (underlined). Following digestion, the product was then ligated into the pETM442 vector (described in (Bagnéris et al., 2008)) to provide an N-terminal 6His-NusA tobacco etch virus (TEV) protease cleavable solubility tag to aid expression and purification. The resulting pETM442-cFLIPS vector was then used as the template for production of pETM442-p22-FLIP which was generated by the introduction

of

a

stop

codon

at

5'tccaaaagagtctcaagtagccttcaaataacttcaggat3'

D196

using

the

primers and

5'atcctgaagttatttgaaggctacttgagactcttttgga3'. GB1-p22-FLIP was produced by subcloning p22-FLIP into the EcoRI and XhoI restriction sites of a modified pET15b vector (kind gift from Prof Paul Driscoll, NIMR) to produce pET15-GB1-p22-FLIP (containing a 6HisGB1 solubility tag). The original pET15b vector was modified by inclusion of the coding sequence for the 56 amino acid B1 domain of E.coli protein G (GB1) 5’ to the multiple cloning site (MCS), the addition of an EcoRI restriction site directly 3’ to GB1 and removal of the MCS EcoRI site. This enabled subcloning of p22-FLIP 3’ to GB1 into the

91

new EcoRI and existing XhoI sites. pETM442-p22-FLIP and pET15-GB1-p22-FLIP were then transformed into BL21(DE3) star® cells (Invitrogen). The cultures were then inoculated into LB medium containing ampicillin (100 µg/ml) at a ratio of 1:100 and grown to an OD600 of between 0.6-0.8. They were subsequently induced with 1mM IPTG and the temperature reduced to 16°C overnight. Cells were harvested by centrifugation and the pellets either frozen at -80°C or resuspended in a buffer comprising 25mM Tris-HCl pH 8.5, 200mM NaCl (buffer A) supplemented with an EDTA free protease inhibitor complex tablet (Roche) and DNase I (10 µg/ml). The purification protocols used for p22-FLIP and GB1-p22-FLIP were near identical. Resuspended pellets from 8L cultures of both proteins were sonicated on ice and the lysates clarified by centrifugation (46,000g for 1 hour). Supernatants were subsequently filtered through a 0.45 µm filter prior to loading on HisTrap FF columns (GE Healthcare) pre-equilibrated with buffer A. These were then washed with buffer B (A and 20mM imidazole) and eluted with buffer C (A and 500mM imidazole). For p22FLIP, the 6His-NusA tag originating from the pETM442-p22-FLIP construct was removed by incubating 6His-NusA-p22-FLIP with TEV protease overnight while dialysing in buffer A (to remove the imidazole) and the solution reapplied to a 5ml HisTrap FF column pre-equilibrated in buffer A. p22-FLIP was eluted from the column using buffer A and 40mM imidazole. Fractions having greater than 95% purity as judged by SDS-PAGE gels were subsequently pooled and frozen at -80°C. Following elution from the 5ml HisTrap column, GB1-p22-FLIP appeared to be over 95% pure. To improve stability, however, GB1-p22-FLIP was subsequently dialysed into a buffer containing 5mM L-arginine, 300mM NaCl, 5mM TCEP and 25mM Imidazole prior to concentration using a 6ml, 10 kDa molecular weight cut-off Vivaspin concentrator (Millipore) and storage at -80 °C.

2.10 Immune complex dephosphorylation Immunoprecipitated complexes were dephosphorylated using the Lambda protein phosphatase (LPP). This is an Mn2+-dependent phosphatase with activity towards phosphorylated serine, threonine and tyrosine residues (Zhuo et al., 1993). Reactions were performed at 30 oC for 30 minutes, in a final volume of 50 µl using varying amounts of LPP (NEB) and Protein MetalloPhosphatase reaction buffer (50mM HEPES pH7.5, 10 mM NaCl, 2 mM DTT, 0.01% Brij35 and 1 mM MnCl2). Subsequently, the reactions were

92

terminated by adding 25 mM Na3VO4 (Gordon, 1991) and 20 mM NaF, followed by cooling the samples on ice.

2.11 Luciferase gene reporter assays Depending on the transfectability of the tested cell types, NF-κB luciferase assays were performed either by transfections or by viral transductions. 70Z/3, MEFs and Jurkat cells were mainly transduced while for HEK293 and HEK293T cells both transductions and transfections were utilised. 2.11.1 NF-κB reporter luciferase assays by transfection One day prior to transfection, 2x105 of HEK293 or 105 of HEK293T cells were seeded in 24-well plates in total volume of 500 µl. Then, cells were transfected with 400 ng of NF-κB luciferase plasmid (pGL.IgK or pGL H2DK), 100 ng of pRL.TK as internal control and 500 ng of pCDNA3 vectors encoding a transactivator gene. Twenty-four hours later the medium was removed, cells were resuspended in 100 µl of medium and transferred into an optical bottom 96-well plate. Luciferase activity was then measured using Dual-Glo® luciferase assay kit (Promega, Madison, WI). First, NF-κB-induced firefly luciferase activity was quantified by adding 50 µl of Dual-Glo luciferase reagent, this signal then was quenched by adding 50 µl of Stop and Glo reagent which also produces stabilised signal from Renilla luciferase. Firefly luciferase activity levels were normalised to that of Renilla which is driven by thymidine kinase promoter of herpes simplex virus and serves as transfection efficiency control. 2.11.2 NF-κB reporter luciferase assays by transduction HEK293/293T, Jurkat, MEFs and 70Z/3 cells were transduced with NF-κB luciferase lentivectors at MOIs of 10, 50, 100 and 500, respectively to generate the stable NF-κB-reporter cell lines. These cells were passaged at least 2-3 times before performing assays. Next, to assess NF-κB activity, 5x104 cells were seeded in optical bottom 96-well plate and transduced with test lentivectors at MOIs of 10, 20, 50 and 100 for HEK293/293T, Jurkat, MEFs and 70Z/3 cells, respectively. Forty-eight hours later, 50 µl BrightGlo® luciferase substrate (Promega) was added to each well and NF-κB-induced luminescence was detected using Varioskan Flash multimode reader (Thermo Scientific).

93

2.12 Statistical analysis Statistical differences between two groups were analysed by the two-tailed unpaired student’s t -test using the GraphPad Prism v4.03 software package (Graphpad Sotware, La Jolla, CA). The calculated P-values are given in the figures. All experiments were performed in triplicates, on at least three independent occasions (unless otherwise stated). The quantitative data are presented as mean±SD.

94

CHAPTER

3 3. The Role of NEMO in IKK Activation by KSHV vFLIP and Cellular FLIPs

95

3.1 Introduction Constitutive activation of NF-κB signalling is linked to cancer development and progression by providing an inflammatory environment for tumour cells and conferring resistance to apoptosis (DiDonato et al., 2012). The oncoproteins KSHV vFLIP and HTLV-1 Tax stably bind to NEMO and constitutively activate the IKK complex, contributing to the neoplastic character of the associated viruses. Cellular FLIPs, which share a high degree of sequence homology with KSHV vFLIP (Figure 1.4), are also strong inducers of the NF-κB cascade (Hu et al., 2000). These proteins become upregulated in a variety of malignancies and are promising targets for cancer therapies (Shirley and Micheau, 2010). Much is known about the role of cFLIPs in the regulation of cell death pathways; however, how they activate NF-κB remains elusive. Potential similarities to vFLIP-induced activation have been suggested, but not investigated. The crystal structure of the vFLIP-NEMO complex has been solved and multiple lines of evidence exist that indicate the necessity of binding to NEMO for the vFLIPmediated induction of IKK (Bagnéris et al., 2008; Field et al., 2003; Liu et al., 2002). It is not clear how this interaction leads to activation. In this chapter, we aimed to explore the role of NEMO in FLIP-induced IKK activation with a particular emphasis on the ubiquitin binding function of NEMO. The UBAN domain of NEMO is indispensable for cytokine-induced IKK activation and has been shown to undergo conformational changes upon binding to linear ubiquitin oligomers (Bloor et al., 2008; Rahighi et al., 2009). Our initial working hypothesis was that vFLIP binding to NEMO may cause similar conformational changes, leading to changes in proximity or conformation of the catalytic IKKα and IKKβ, in favour of the activation. Therefore, we first set out to determine whether FLIP proteins required ubiquitination and the ubiquitin binding activities of the NEMO to stimulate IKK, or whether binding to NEMO was sufficient to bypass these processes. In the following sections, I provide a brief description of the key IKK-dependent functions of the NEMO and also, a summary of the current knowledge on the interaction of NEMO with vFLIP, Tax and cellular FLIPs.

96

3.1.1 Molecular control of IKKs by NEMO

3.1.1.1 NEMO mediates the Ub-dependent activation of IKK Activation of the canonical NF-κB pathway is contingent upon non-degradative polyubiquitination. NEMO plays a crucial role in ubiquitin-mediated activation of the IKK complex as it specifically recognises polyubiquitin chains through its UBAN domain and also, becomes ubiquitinated itself (Tokunaga et al., 2009; Wu et al., 2006a). Cells expressing NEMO mutants defective in Ub-binding exhibit impaired activation of IKKα/β in response to IL-1 and TNFα (Ea et al., 2006; Windheim et al., 2008; Wu et al., 2006a). Importantly, similar mutations of the UBAN domain of NEMO are found in patients afflicted with EDA-ID and IP, emphasising the essential role of this domain in vivo (Hubeau et al., 2011; Smahi et al., 2000). Initially it was thought that the UBAN specifically binds to K63-linked ubiquitin chains (Wu et al., 2006a); later studies, however, demonstrated that it displays 100-fold higher affinity for linearly-linked ubiquitin chains (Rahighi et al., 2009; Xu et al., 2011). Consistent with these findings, full-length NEMO mutants engineered to selectively bind to K63-linked polyubiquitin only weakly activate NF-κB (Kensche et al., 2012). One further study, however, demonstrated that despite the much higher preference of NEMO C-terminus for linear ubiquitin oligomers, its immobilisation – which may be reminiscent of cellular IKK oligomerisation - enhances its affinity towards K63 linkages (Hadian et al., 2011). Therefore, in cells, both types of these linkages may function in a certain spatiotemporal order to achieve the maximal levels of NF-κB activation. As previously mentioned, activation of the IKKs seems to occur through an induced proximity which results from dense organisation of the signalling components or structural changes induced by binding of adapter proteins. NEMO-mediated ubiquitin binding has been proposed to participate in both processes. Upon receptor-mediated activation of the NF-κB, K63-linked ubiquitin chains synthesised by E3 ligases such as TRAF2/5 and cIAP1/2 (in TNFα signalling), and TRAF6 (in IL-1 signalling), provide a recruitment platform for TAB2/TAK1 and NEMO/IKK complex. These K63 ubiquitin chains also recruit LUBAC (Emmerich et al., 2013) to generate linear polyubiquitin oligomers on the NEMO and other substrates such as RIP1 (Gerlach et al., 2011), resulting in further stabilisation of the signalling complex. UBAN-mediated attachment to ubiquitin chains is suspected to enable NEMO 97

to facilitate co-localisation of IKKs with TAK1, leading to phosphorylation and activation of the IKKα/β (Clark et al., 2013). Alternatively, binding of NEMO to ubiquitin have been suggested to cause IKK activation via other mechanisms. Crystal structure studies by Rahighi et al., showed that binding of linear di-ubiquitin molecules to the UBAN domain, induces an straightening of this coil-coiled region (Rahighi et al., 2009). It was proposed that this conformational change may then transmit to the N-terminal regions of NEMO, resulting in the reorientation of the catalytic IKKs, favouring activation by autophosphorylation. Relevant to this notion, site-specific mutations can be introduced to NEMO (such as K270A in murine NEMO) which overcome the need for ubiquitin binding and constitutively activate IKK in the absence of any proinflammatory stimuli (Bloor et al., 2008). Such mutations may mimic the conformational alterations induced by the binding of the ubiquitin oligomers. That said, whether structural changes to NEMO translate to allosteric effects on IKKα/β, remains to be demonstrated.

3.1.1.2 Post-translational modifications of NEMO fine-tune the IKK activity In addition to ubiquitination, other post-translational modifications such as phosphorylation and SUMOylation, also play crucial roles in modulating the function of NEMO (Perkins, 2006). Following NF-κB stimulation, IKKβ phosphorylates NEMO at several amino acids, located near the N-terminus and the C-terminus (Carter et al., 2003; Palkowitsch et al., 2008; Prajapati and Gaynor, 2002). Notably, phosphorylation of serine68 of NEMO, which resides within the IKKα/β-binding domain, disrupts the NEMO dimerisation and NEMO-IKKβ interaction (Palkowitsch et al., 2008). Accumulating evidence suggest that this modification together with the simultaneous authophosphorylation of IKKβ at its C-terminal NBD (Delhase, 1999; Schomer-Miller et al., 2006), serve as an intrinsic regulatory mechanism preventing hyperactivation of the IKKα/β following stimulation. SUMOylation of NEMO is another important post-translational modification, found to be critical for its role in the nuclear-initiated NF-κB activation (Miyamoto, 2011). In cells, a subset of NEMO exists as an unbound form that shuttles between nucleus and cytoplasm. This form of NEMO appears to be essential for connecting the nuclear DNA damage events to cytoplasmic activation of IKK. Upon genotoxic stress, nuclear NEMO becomes SUMOylated by the SUMO E3 ligase, PIASy (protein inhibitor of activated 98

STAT y) at the lysines 277 and 309 (Huang et al., 2003; Mabb et al., 2006; Stilmann et al., 2009). This marks NEMO for cIAP1-mediated monoubiquitination at the same lysine residues, leading to its nuclear export and subsequent activation of TAK1, IKK and NFκB, in a manner that requires ELKS (Huang et al., 2003; Wu et al., 2006b). Interestingly, SUMO-fused mutants of NEMO localise in the nucleus, indicating the importance of SUMOylation in determining the subcellular localisation of NEMO (McCool and Miyamoto, 2012). Taken together, these findings underscore the significance of the posttranslational modifications of NEMO in producing a highly regulated and signal-specific NF-κB signalling.

3.1.1.3 NEMO ensures the substrate-specificity of the IKKβ Most studies on NEMO have focused on its well-known role in the activation of IKKα/β. Nevertheless, a recent study showed that constitutively active IKKβ fails to activate NF-κB in the absence of NEMO, suggesting that NEMO may have additional roles (Schröfelbauer et al., 2012). Further investigations by Schrofelbauer et al., revealed that NEMO directs the substrate-specificity of IKKβ towards IκBα, through its ZF domain. Not surprisingly, various in-frame deletions and point mutations of the ZF domain are associated with the disease, EDA-ID (Cordier et al., 2008). IKKα and IKKβ are pleiotropic enzymes which target a broad spectrum of NF-κBrelated (e.g. IκBs, p65, and p105) and NF-κB-unrelated substrates (e.g. β-catenin, FOXO3 and p53) (Hinz and Scheidereit, 2014). Mutations of NEMO that fail to recruit IκBα, result in an increased phosphorylation of p65 and p105 (Schröfelbauer et al., 2012). This raises the possibility that NEMO may use distinct regions to target the activity of catalytic IKKs to other substrates. In support of this notion, panr2 mice that harbour a L153P mutation in NEMO exhibit severe defects in phosphorylation of p65 and p105, despite normal phosphorylation and degradation of IκBα (Siggs et al., 2010). Taken together, these findings indicate that NEMO is necessary not only for the activation of the IKK complex but also to ensure the substrate-selectivity of the enzymatic subunits.

3.1.1.4 NEMO serves as a docking site for various IKK-regulating proteins Activation of IKK is a transient event which depends on dynamic regulatory processes to produce an optimal signal. The balance towards activation or deactivation is determined by selective recruitment of the upstream co-factors to the IKK complex machinery. In this regard, the elongated coiled-coil structure of NEMO provides a 99

docking platform for numerous regulatory proteins. Tables 3.1 gives a summary of such proteins which directly or through an adapter protein, interact with NEMO and modulate IKK activity. Apart from the functions mentioned here, NEMO regulates a plethora of other NF-κB dependent (e.g. interferon production) and NF-κB independent mechanisms which have been reviewed in (Clark et al., 2013). Table 3.1. NEMO-interacting proteins. Interactor

Interaction region

Function Reference

A20

aa 95-218

inh

(Zhang et al., 2000)

ABIN1

aa 50-100

inh

(Mauro et al., 2006)

ABIN2

aa 174-306

inh

(Liu et al., 2004)

aa 1-150

inh

(Li et al., 2000; Mauro et al., 2003;

ActI/CIKS

Qian et al., 2004) ATM

n.d

act

(Wu et al., 2006b)

CARMA1,3

n.d

act

(Stilo et al., 2004)

CBP

n.d

inh

(Verma et al., 2004)

CSN3

aa 297-419

inh

(Hong et al., 2001; Rual et al., 2005)

CYLD

ZF

inh

(Kovalenko et al., 2003; Saito et al., 2004; Trompouki et al., 2003)

EHV2 vCLAP

aa 300-419

act

(Poyet et al., 2001)

n.d

act

(Ducut Sigala et al., 2004)

HIF-2α

aa 50-358

act

(Bracken et al., 2005)

HSP70

aa 25-320

inh

(Ran et al., 2004)

Htt

aa 1-134

act

(Khoshnan et al., 2004)

NLRX1

UBAN

inh

(Xia et al., 2011)

inh

(Shibata et al., 2012)

PAN1

K63 and linear Ub chains bound to NEMO n.d

inh

(Bruey et al., 2004)

PARP1

aa 1-126

act

(Stilmann et al., 2009)

PIASy

aa 1-120

act

(Huang et al., 2003; Mabb et al., 2006)

PIDD

n.d

act

(Janssens et al., 2005)

PP2A

aa 121-179

act

(Kray et al., 2005)

PP2Cβ

n.d

inh

(Prajapati et al., 2004)

ELKS

p47

100

RIP1

UBAN

act

(Wu et al., 2006a; Zhang et al., 2000)

SRC3

n.d

act

(Amazit et al., 2007)

TANK

aa 200-250 (mNEMO)

act

(Bonif et al., 2006; Chariot et al., 2002)

TFG

LZ

act

(Miranda et al., 2006)

TRAF1/2

n.d

act

(Devin et al., 2001; Henn et al., 2007)

TRUSS

n.d

act

(Soond et al., 2003)

UBAN

act

(Bloor et al., 2008; Ea et al., 2006; Wu

Ubiquitin

et al., 2006a) ZNF216

n.d

act

(Huang et al., 2004a)

n.d: not determined, inh: inhibition, act: activation, mNEMO: murine NEMO. 3.1.2 Interactions of the Tax, vFLIP and cFLIPs with NEMO It is now well-established that the persistent activation of the canonical IKKs by Tax and vFLIP requires stable assembly with NEMO (Chu et al., 1999; Field et al., 2003; Harhaj and Sun, 1999). Although the mechanism by which these interactions lead to NFκB activation is still a matter of mystery (See 5.1 for the suggested mechanisms). Using coimmunoprecipitation assays, Tax has been shown to interact with two distinct but highly homologous motifs in NEMO: one in the N-terminus (aa 100-140) and the other, near the C-terminus (aa 312-340) (Figure 3.1A) (Xiao et al., 2000). The region of Tax responsible for binding to NEMO is localised to a leucine rich repeat (LRR) motif, containing amino acid residues 105-141. Site-specific mutations of the LRR domain which disrupt binding to NEMO, such as those in the Tax M22 mutant (T130A, L131S), abrogate Tax-induced IKK activation. Strikingly, in-frame fusion of the M22 to NEMO restores its NF-κB activating function, further emphasising the necessity of interaction with NEMO for Tax-mediated IKK activation (Xiao and Sun, 2000; Xiao et al., 2000). vFLIP, despite employing a similar mode of IKK activation which depends on physical interaction with NEMO, appears to target a different region of NEMO. Studies conducted in our laboratory have mapped the vFLIP-binding region of NEMO to its HLX-2 domain (residues 192-252) (Bagnéris et al., 2008; Field et al., 2003). Subsequent crystal structure studies determined that this region of NEMO forms a parallel intermolecular coiled -coil recognised by two vFLIP molecules that interact through their respective DED1 motifs (Fig 3.1B and C) (Bagnéris et al., 2008). Interactions of the NEMO helices with DED1 is facilitated by the presence of two deep adjacent clefts, of 101

which, cleft1 mediates the majority of the interprotein contacts (Fig 3.1C). Mutating the key amino acids of the cleft1-HLX2 interface (e.g. A57L in vFLIP and D242R in NEMO) results in complete abolition of the complex formation. While some homologues of the KSHV vFLIP (e.g. p22-cFLIP) are similarly potent activators of the IKK, others fail to stimulate this complex (e.g. MC159L). This raises the possibility that the ability of FLIPs to activate IKK may depend on the conservation of structural properties affecting the FLIP-NEMO interface. In support of this concept, structural alignment of the KSHV vFLIP and MC159L suggested that the sequence differences in the latter, result in obscuring cleft2 and complete closure of cleft1, thereby abrogating the NEMO binding (Figure 3.1D). In the case of p22-FLIP, however, there are no amino acid substitutions that would lead to major steric clashes and hence drastic remodelling of cleft1 and cleft2. Therefore, FLIP-mediated activation of the canonical NF-κB pathway may strongly rely on the preservation of the NEMO contacts mediated by clefts of the DED1 motif (Figure 3.1D)(Bagnéris et al., 2008). Among cFLIP variants, the proteolytic fragments p43-FLIP and p22-FLIP, have been reported to directly interact with NEMO (Golks et al., 2006; Neumann et al., 2010). p43-FLIP is generated by procapase-8 cleavage of the cFLIPL at the DISC following CD95 stimulation (Neumann et al., 2010), however, p22-FLIP can be catalysed from all isoforms (cFLIPL/S/R) and differs form p43-FLIP in that it can be also produced in nonapoptotic cells. Therefore, p22-FLIP is believed to be the final cleavage product of cFLIP proteins, serving as the mediator of NF-κB activation, in a manner similar to that of KSHV vFLIP (Bagnéris et al., 2008; Golks et al., 2006). Nevertheless, there exists no experimental data demonstrating whether p22-FLIP (or any other cFLIP variant) targets the same vFLIP-binding region of the NEMO and whether this possible interaction is necessary for cFLIP-induced IKK activation.

102

Figure 3.1. Structure of NEMO and its complexes. A) Domain organisation of the NEMO and its interaction sites with IKKα/β, vFLIP, ubiquitin and Tax. CC: colied coil, HLX: helical domain, LZ: leucine zipper, ZF: zinc finger. B) Crystal structures of NEMO fragments. Three-dimensional structure of the full-length NEMO is not available 103

yet; however, the crystal structures of several NEMO fragments have been resolved. These include: N-terminal kinase binding domain of NEMO in complex with IKKβ (Rushe et al., 2008), HLX2 in complex with KSHV vFLIP (Bagnéris et al., 2008), CC2-LZ fragment in complex with linear di-ubiquitin (Rahighi et al., 2009) and ZF domain (Cordier et al., 2008). C) The KSHV vFLIP-NEMO interface. Left panel: KSHV vFLIP depicted as green surface contains two adjacent vertical clefts in its DED1 motif. Cleft1 and cleft2, highlighted in magenta and yellow, respectively, are found to form a complementary binding surface for NEMO helices. Right panel: two vFLIP molecules are seen in complex with a central region of two parallel alpha helices of NEMO. Both clefts of the vFLIP are pivotal to the recognition of the NEMO dimer, although, the majority of interprotein contacts are mediated by cleft1. D) Structural comparisons of KSHV vFLIP, MC159 vFLIP and p22-FLIP. Analysis of the electrostatic potential surfaces of these proteins shows that the cleft1 and cleft2 are absent in MC159 vFLIP. However, homology model of p22-FLIP suggests that two binding clefts are conserved in this protein and that the analogous interactions to those observed with KSHV vFLIP are possible with NEMO. Reproduced from (Zheng et al., 2011) and (Bagnéris et al., 2008) with a permission from the publishers.

3.2 Aims of the chapter •

Identify the domains of NEMO that are required for IKK activation by vFLIP and cellular FLIPs

•

Examine whether cellular FLIPs physically interact with NEMO and if they do, which region of the NEMO is targeted

•

Examine the importance of linear ubiquitination in FLIP-induced activation of IKK

104

3.3 Results 3.3.1 Generation of a NF-κB reporter luciferase assay system In order to establish a cell-based NF-κB activation assay, we generated a lentiviral vector (LV) that allows for NF-κB-dependent transcription of the firefly luciferase under the control of the minimal CMV promoter and four repeats of an NF-κB binding site (Figure 3.2A). First, to test the specificity of the system, we transduced cells with the lentivectors encoding GFP, or vFLIP, or NF-κB luciferase with and without vFLIP or GFP (each MOI=10). Forty-eight hours later, cells were lysed and activity of the firefly luciferase was measured using Bright-Glo detection system. As shown in Figure 3.2B, the IKK activator vFLIP robustly induced expression of luciferase by more than 100-fold increase (compared to unstimulated cells), whereas the GFP control failed to induce any luciferase activity. Next, to examine the sensitivity of our reporter system, we transduced HEK293T cells with increasing MOIs (0.5, 1, 2, 4, 8, 16 and 32) of both NF-κB luciferase and vFLIP lentivectors. Figure 3.3C and D show that the levels of luciferase activity correlated well with MOIs of the vFLIP LV, except in the condition where MOI of 32 was used for the NF-κB luciferase LV. Since the transductions of vFLIP and NF-κB luciferase were performed simultaneously, this could be due to a competition in viral entry between the two LVs at high MOIs. In all the luciferase experiments discussed later, I first transduced cells with NF-κB luciferase LV to produce a reporter cell line and then, activation assays were performed after at least two cell passages. Collectively, the data presented in Figure 3.2 show that our LV-based NF-κB activation assay is highly sensitive and specific. An important advantage of this LV-based system is that it allows for genetic engineering of poorly transfectable cells (e.g. MEFs) that cannot be assessed using conventional reporter assays based on transient transfection. Furthermore, working with stable reporter cells can circumvent the problem with variations in transfection efficiency that may mask the NF-κB-dependent signal.

105

Figure 3.2. Generation of a cell-based NF-κB reporter luciferase assay. A) Schematic representation of the lentiviral construct (pSIN NF-κB Luc) used for the NFκB reporter luciferase assays. CMV MP: minimal promoter of cytomegalovirus, cPPT: central polypurine tract, LTR: long terminal repeat, NF-κB RE: NF-κB response element, RRE: rev response element, SIN: self-inactivating, WPRE: woodchuck post-transcription regulatory element. B) NF-κB reporter luciferase assay to check the specificity of the 106

system. HEK293T cells, seeded in 96-well plates (2x104/well) were transduced with lentiviral vectors encoding GFP, or vFLIP, or NF-κB luciferase with and without GFP or vFLIP. The NF-κB dependent luciferase activity was measured 48 hours posttransduction. C) NF-κB reporter luciferase assay to check the sensitivity of the system. The same number of HEK293T cells, as described in (B), were transduced with LVs encoding NF-κB reporter luciferase and vFLIP over a broad range of MOI (0.5, 1, 2, 4, 8, 16, 32) for each lentivector. Cells were lysed 48 hours following transduction and the luciferase activities were measured using Bright GLO luminescence detection kit (Promega). The results are represented as RLU (relative luminesce units). D) as in (C), but the results have been represented as normalised fold change values, calculated by dividing of the RLU of activated cells to RLU of the control cells.

3.3.2 Unlike vFLIP and Tax, cellular FLIPs require Ub-binding function of NEMO to activate NF-κB signalling To determine which functions of NEMO were necessary for IKK activation by the FLIP proteins, we used the NEMO null cell line 1.3E2 - a derivative of the mouse preB cell line 70Z/3- harbouring an NF-κB responsive luciferase gene. These cells were reconstituted with wild-type NEMO, a point mutant which does not bind linear ubiquitin (F312A, (Rahighi et al., 2009)), or a F238R/D242R mutant suggested by structural studies to disrupt vFLIP binding (Bagnéris et al., 2008). To do this, we used lentiviral vectors expressing wild type or mutant NEMO together with an mCherry fluorescent protein (See Figure 2.1A). Cells were transduced so that approximately 50% were mCherry positive. We then isolated cell clones, expanded those that were mCherry positive and performed immunoblotting to establish that the transduced cells expressed NEMO (Figure 3.3A). Cells were then infected with a second lentiviral vector expressing cFLIP variants (p22FLIP, FLIPS, FLIPL), or vFLIP, or Tax together with eGFP. After 48 hrs eGFP was monitored and cells with transduction rates over 80% were used to measure luciferase activity. Figure 3.3C shows that both vFLIP and Tax could activate IKK in the absence of its ubiquitin binding function, unlike the control Toll-like receptor agonist lipopolysaccharide (Figure 3.3B). In contrast, cFLIPL, cFLIPS, and their proteolytic product p22-FLIP all require the ubiquitin binding function of NEMO to activate IKK (Figure 3.3D). To our surprise, all cFLIP variants were able to activate vFLIP-binding 107

deficient NEMO, suggesting that cellular FLIPs either target different motifs of NEMO, or bind to it indirectly, or function at some upstream step (Figure 3.3D). Activation of F238/D242RR mutant by Tax, which binds to distinct regions of NEMO, and also LPS further showed that this mutation specifically inhibits vFLIP and does not impair other activatory mechanisms (Figure 3.3C). As previously mentioned, the Ala57 residue of vFLIP plays an important role in binding to NEMO and an A57L mutation reportedly impairs NF-κB inducing activity of the protein. The LAE sequence harbouring this alanine is conserved in both KSHV and cellular FLIPs (marked with * in Figure 3.4A). To test if this residue was also important for the function of cFLIPs, I generated equivalent A56L mutants of all three isoforms and performed an NF-κB activation assay on 293T cells. The mutation resulted in almost complete inactivation of all FLIPs (Figure 3.4C). However, western blot analysis showed low levels of protein expression for vFLIP, p22-FLIP and FLIPS, indicating that this mutation renders the proteins unstable (Figure 3.4B). This observation makes the relevant results inconclusive; nevertheless, similar protein levels in the case of WT and mutant FLIPL suggests that the conserved L57 residue may play a role in NF-κB induction by cFLIP variants.

108

Figure 3.3. Mutational studies on NEMO reveal differential NF-κB activation mechanisms for the KSHV vFLIP and cellular FLIP isoforms. A) Immunoblot showing the expression of NEMO in 1.3E2 cells, 70Z/3 cells and 1.3E2 cells reconstituted with wild-type or mutant NEMO. The blot was re-probed for GAPDH to ensure even protein loading. B) To generate stable NF-κB reporter cell lines, cells were transduced with lentiviral vectors encoding NF-κB responsive luciferase gene (MOI=500). The transduced cells were passaged multiple times prior to NF-κB luciferase assays. Subsequently, the luciferase reporter assays were performed 6 hours after stimulation with LPS (10 µg/ml) or C) 48 hours following transduction with lentivectors encoding vFLIP, Tax (MOI=50) and D) cFLIP variants (MOI=100). Data are representative of three independent experiments performed in triplicates. Error bars indicate SD of the mean values.

109

Figure 3.4. NF-κB activation by A57L vFLIP and A56L cFLIP isoforms.

A)

Sequence alignment of the KSHV vFLIP and cFLIP. Amino acid sequences of the proteins were retrieved from NCBI protein database and aligned using PRALINE multiple sequence alignment software. Accession numbers of the KSHV vFLIP and cFLIPS are AAD46498.1 and NP_001120656.1, respectively. Amino acid residues of vFLIP predicted by structural studies to be important for vFLIP-NEMO interaction have 110

been shown inside yellow circles. The equivalent conserved residues of cFLIP are depicted inside blue circles. Ala57 of KSHV vFLIP (indicated by *) which is essential for binding to NEMO and its adjacent amino acids (56-58, LAE) are highly conserved in cellular FLIPs. B) HEK293T were seeded in 24-well plates cells (2x105/well) 24 hours prior to transfection. Each well was transfected with 400 ng NF-κB firefly luciferase plasmid (pGL.IgK), 100ng Renilla luciferase plasmid (pRL.TK) as internal control and 300ng of pCDNA3 vectors encoding WT, or A57L vFLIP, or A56L cFLIP isoforms. NFκB induction levels were measured after 24 hours using Dual-GLO kit (Promega). B) Western blot showing the expression of the WT and mutants versions of the vFLIP and cFLIP variants. The mutations severely lower the stability of both viral and cellular FLIPs. Quantitative data are representative of three independent experiments, performed in triplicates. 3.3.3 cFLIPs generate an active IKK without stable interaction with NEMO Activation of the vFLIP-binding deficient NEMO by cFLIPs suggested that these proteins may target a distinct domain of NEMO or function upstream of this protein. Therefore, we next sought to determine whether cFLIP isoforms associate with NEMO and generate a constitutively active IKK. To do this, we transfected HEK293T cells with expression plasmids for FLIPS, FLIPL, a non-cleavable version of cFLIPL (cFLIPL D196E/D376N) and p22-FLIP as well as vFLIP as a positive control for IKK direct interaction. We then immunoprecipitated the IKK complex using an anti-NEMO antibody and measured its ability to phosphorylate recombinant IκB. As can be seen in Figure 3.5 panel A, in each case the cFLIP isoforms generated an activated IKK. The level of activation was comparable or greater than that generated by vFLIP, though less than the transient activation observed following TNFα treatment. We then examined whether the cFLIP isoforms were found associated with the activated kinase and found no evidence for the presence of cFLIP isoforms in the IKK immune-precipitates, though vFLIP was clearly present as revealed by both anti-NEMO and anti-FLIP immunoprecipitations (Figure 3.5B). This is in contrast to the previous reports that showed p22-FLIP and p43-FLIP fragments stably associated with IKK (Golks et al., 2006; Neumann et al., 2010). Clearly, differences in the experimental conditions could explain these differing results; however, our data demonstrate that the cFLIP isoforms can generate an active IKK without remaining physically associated with the IKK complex.

111

Figure 3.5. Cellular FLIPs constitutively activate IKK complex without stable association with NEMO. A) In vitro kinase assays to determine the activation of IKK. HEK293T cells seeded in 6-well plates (5x105 cells/well) were transfected with pCDNA3 vectors encoding p22FLIP, FLIPS, FLIPL, non-cleavable FLIPL, vFLIP or Tax. Fortyeight hours later, whole cell lysates were extracted and subjected to immunoprecipitation (IP) with anti-NEMO or immunoblotting (IB) with the indicated antibodies. In vitro kinase assay was then performed on immuno-isolated IKK complexes using GST-IκBα (1-54) and γ32P-ATP as substrates. Extract of cells treated with TNFα (10ng/ml) for 5 min was used as a positive control. Relative band intensity of phosphorylated GST-IκB was quantified by ImageQuant TL Plus7.0 and normalised to the corresponding immunoprecipitated IKKα/β levels. B) cFLIP variants are not found in complex with NEMO. HEK293T cells seeded in 15 cm2 dishes were transfected with pCDNA3 constucts, using FuGene HD transfection reagent. Cells were lysed 48 hrs later and the 112

extracts were immunoprecipitated with 2 µg of antibodies against NEMO, cFLIP and vFLIP, or their isotype-matched controls. The immunoprecipitates were subject to 10% SDS-PAGE and were analysed by western blotting. Data shown are representative of at least four independent repeats. * indicates the position of p43-FLIP bands. cF: cFLIP, I: isotype control, N: NEMO, ns: non-specific, vF: vFLIP. 3.3.4 cFLIPL requires LUBAC to activate IKK As cFLIP isoforms required the ubiquitin binding function of NEMO to activate IKK (Figure 3.3), we examined the effect of ubiquitination pathways on their action. Linear ubiquitin chain binding to NEMO is crucial for IKK activation by TNFα and these are generated by the trimeric LUBAC, composed of HOIL-1, HOIP and SHARPIN (Iwai et al., 2014). We therefore generated HEK293 cells that were stably transduced with lentiviral vectors containing short hairpin RNAs targeting HOIL-1, HOIP and SHARPIN (Figure 3.6A). Cells transduced with scramble non-targeting shRNA served as control. As previously reported LUBAC was required for IKK activation by TNFα (Figure 3.6B) but was dispensable for activation by vFLIP (Tolani et al., 2014) and Tax (Figure 3.6C). A clear blockade of cFLIPL and nc-cFLIPL induced activation of IKK was observed in the LUBAC knock-down cells. However, cFLIPS and p22-FLIP were unaffected by LUBAC knock-down (Figure 3.6C). Lack of inhibition in the LUBAC KD cells, led us to hypothesise that cFLIPS and p22-FLIP may specifically require binding of K63-linked ubiquitin chains to NEMO. CYLD is a deubiquitinase (Komander et al., 2009) which can remove ubiquitin chains with K63 as well as linear linkages and has been shown to target several components of IKK signalling such as NEMO, RIP1, TRAF2 and TRAF6, resulting in inhibition of NFκB signalling (See 1.2.3.3). Therefore, we overexpressed CYLD in our scramble and LUBAC KD HEK293 cells (Figure 3.7A). Upregulation of CYLD caused a significant decrease in TNFα-induced NF-κB activation in LUBAC KD cells, but not control cells (Figure 3.7B). As seen in Figure 3.7C, cFLIPL was once again inhibited in the CYLD overexpressed cells, while p22-FLIP, cFLIPS and vFLIP were unaffected. Nevertheless, in the case of cFLIPS-induced activation, a statistically significant but not considerable decrease could be detected in the CYLD transduced/LUBAC KD cells when compared to control cells. The lack of requirement for ubiquitin signalling by cFLIPS and p22-FLIP required further investigation as the ubiquitin binding function of NEMO was clearly required. This conundrum will be addressed and discussed later in this thesis.

113

Figure 3.6. Effects of knocking down LUBAC on the NF-κB activation ability of the Tax, vFLIP and cellular FLIPs. A) HEK293 cells were transduced with lentiviral vectors encoding either scramble control shRNA (scr) or shRNAs targeting HOIL-1, HOIP and SHARPIN, sequentially. Generation of LUBAC KD cells was confirmed with western blotting against each component of the complex and the housekeeping protein, GAPDH. ns; non-specific B) TNFα-induced NF-κB activation which was measured 6 hours after stimulation at 10 ng/ml concentration is inhibited in LUBAC KD cells. C) Subconfluent monolayers of scramble and LUBAC KD HEK293 cells seeded in 24 wellplates (2x105/well) were co-transfected with an NF-κB−firefly Luc reporter construct (300ng/well) and a Renilla Luc reporter vector (normalisation control,100ng/well) along with an empty or transactivator expressing pCDNA3 vectors (500ng/well). The luciferase reporter assay was performed 24 hrs post-transfection as described in “section 2.11.1”. Values shown are the mean±SD of fold changes in the luciferase activity, from one representative experiment out of three, performed in triplicates; ** denotes p